FORM 2
THE PATENTS ACT, 1970
(39 OF 1970)
&
The Patents Rules, 2003
COMPLETE SPECIFICATION
(See section 10; rule 13)
1. Title of the invention – “AN INVESTIGATION AND ANALYSIS OF MACHINE
LEARNING TECHNIQUES TO MEASURE STUDENTS’
PERFORMANCE.”
2. Applicant(s)
a) NAME: RK UNIVERSITY
b) NATIONALITY: An Indian Education Institute
c) ADDRESS: RK University, Bhavnagar Highway, Tramba, Rajkot, Gujarat, 360020,
INDIA
3. PREAMBLE TO THE DESCRIPTION:
The following specification particularly describes the invention and the manner in
which it is to be performed.
Page 2 of 23
FIELD OF THE INVENTION
The present invention relates to the students’ performance in education institutes via
machine learning. More particularly present invention relates to the field of machine
5 learning technology. More particularly, the present invention relates to a machine
learning based system for predicting the academic performance of students.
BACKGROUND OF THE INVENTION
10 Universities are always trying to get a good rating from the accrediting bodies and for this
purpose universities are diversifying the means they need to attract international students.
It is necessary to evaluate the academic performance of international students in order to
improve university management.
15 The machine learning technique can be applied to the above problem as it has the potential
to classify the students based on their academic performance, thus enabling the
educational institutions to predict the required students for their universities.
There are several reasons why there is a need to assess the academic performance of
20 international students, for the following reasons: students come from diverse ethnic and
academic backgrounds, assessing their academic performance helps universities plan
timely corrective actions, students with Better academic performance helps the university
to grow globally, and the proportion of international students is an important parameter
when evaluating university rankings.
25
Considering all these reasons, it is necessary for universities to predict the academic
performance of international students.
Considering the previous discussion, it becomes clear that a machine learning based
30 system for predicting the academic performance of international students is needed.
US 20150050637 A1 disclose the system and method for assessing student performance.
The method includes receiving student data, third party data, a set of predetermined
Page 3 of 23
thresholds, and an activity guide at a computer processor. The student data and third party
data are linked into a collected data, at least a part of which is evaluated against the set of
predetermined thresholds. The student's performance is classified based on a
determination if the collected data meets the predetermined thresholds. The collected data
5 and classifications are stored and displayed, and a performance score is provided. A
targeted strategy is provided for the student by identifying a set of targeted actions
corresponding to the student classifications, and displaying targeted actions to authorized
users.
10 US 7493077 B2 disclose the system and method enable processing and displaying of test
results in accordance with information and specifications provided by the client requiring
such services. Skills examined by a test are identified, and skill-by-skill analysis of test
results is provided. Performance within constituent skill categories are compared to
applicable standards, thereby providing criterion-referenced conclusions about a student's
15 performance within each skill category tested. Evaluation and recommendation
statements are generated for students or groups of students based on their test performance
and procedures for automatically generating such statements are provided. Test results
are also used to identify skills in which individual students and groups have the greatest
potential for growth. Test reports are produced in print and electronically using the same
20 electronic document structures and data source files to ensure consistency between the
two display mechanisms. The analyses and reports generated provide instructional
information tailored to the student's or group's needs, as identified by analysis of test
performance. Teachers and administrators can track the progress of students and groups
through useful, accurate, and easily accessible test reports.
25
SUMMARY OF THE INVENTION
The present disclosure relates to a machine learning-based system for predicting the
academic performance of international students. The present disclosure proposes a system
30 for predicting the academic performance of international students studying at a university
in North India, where the prediction is made through the use of machine learning
techniques and, after classification, a human-interpretable explanation of the classified
Page 4 of 23
results is obtained using LIME .Classification is done based on the student's
demographics and their participation in academic activities, using attributes such as
attendance percentage, pending repeats, economic level and geographic region, etc. to
develop a model to predict a student's performance as fair or poor. Machine learning
5 techniques such as Logistic Regression, Naive Bayes, CART and Random Forest are used
in the proposed system for prediction. The proposed system will be evaluated using
metrics such as classification accuracy, sensitivity, specificity and area under the ROC
curve. The experiments are performed with the open source statistical program R. The
evaluation results show that the classification accuracy is over 90% during the
10 experiment. The attributes or characteristics such as attendance percentage and pending
reappearance have a major impact on the prediction results. The present disclosure aims
to provide a machine learning based system for predicting the academic performance of
international students. The system includes: a data pre-processing unit to prepare the final
data set of international students, the initial data set consisted of international students
15 and the final data set after pre-processing consisted of students with a mix of
undergraduate and graduate students from different countries a feature extraction
processing unit for extracting the features that are really important and removing all
redundant and unimportant features; a classification processing unit for classifying the
students based on their academic performance using four different classification
20 algorithms, namely logistic regression, naïve Bayes, CARTs and random forests, to
perform the classification; an evaluation processing unit for performing an evaluation of
the academic performance prediction system by calculating the metrics, namely
classification accuracy, sensitivity, specificity and area under the ROC curve; and a result
interpretation processing unit for providing an explanation behind the results seen
25 interpretable by humans using an approach called LIME to obtain said explanation. A
goal of the present disclosure is to provide a machine learning based system for predicting
the academic performance of international students.
Another aim of the present disclosure is to consider the real data set instead of the standard
30 data set for a better prediction, the data set being from international students studying at
a north Indian university. Another subject of the present disclosure is the use of logistic
regression, naive bayes, CART and random forests as machine learning methods to
Page 5 of 23
perform the prediction. Another objective of the present disclosure is to evaluate the
various combinations of features in order to select the right features according to their
predictive ability. Another goal of the present disclosure is to obtain an interpretable
human explanation of the results using LIME, a recently proposed model-agnostic
5 approach. Another objective of the present disclosure is to enable universities to predict
the academic performance of the interacting students so that universities can plan timely
corrective management actions.
These and other features of the ideas provided herein will become more obvious to the
10 ones of talent inside the art in view of the accompanying drawings and following
description, which divulge unique embodiments of such principles in more detail.
DETAILED DESCRIPTION OF THE INVENTION
15 The detailed description set forth below in connection with the appended drawings is
intended as a description of certain embodiments of a collapsible wheeled walker
apparatus and related method of use and is not intended to represent the only forms that
may be developed or utilized. The description sets forth the various structure and/or
functions in connection with the illustrated embodiments, but it is to be understood,
20 however, that the same or equivalent structure and/or functions may be accomplished by
different embodiments that are also intended to be encompassed within the scope of the
present disclosure. It is further understood that the use of relational terms such as first and
second, and the like are used solely to distinguish one entity from another without
necessarily requiring or implying any actual such relationship or order between such
25 entities.
The background, summary and the above description includes information that may be
useful in understanding the present disclosure. It is not an admission that any of the
information provided herein is prior art or relevant to the presently claimed inventive
30 subject matter, or that any publication specifically or implicitly referenced is prior art.
In some embodiments, the numbers expressing dimensions, quantities, quantiles of
Page 6 of 23
ingredients, properties of materials, and so forth, used to describe and claim certain
embodiments of the disclosure are to be understood as being modified in some instances
by the term “about.” Accordingly, in some embodiments, the numerical parameters set
forth in the written description and attached claims are approximations that can vary
5 depending upon the desired properties sought to be obtained by a particular embodiment.
In some embodiments, the numerical parameters should be construed in light of the
number of reported significant digits and by applying ordinary rounding techniques.
Notwithstanding that the numerical ranges and parameters setting forth the broad scope
of some embodiments of the disclosure are approximations, the numerical values set forth
10 in the specific examples are reported as precisely as practicable. The numerical values
presented in some embodiments of the disclose may contain certain errors necessarily
resulting from the standard deviation found in their respective testing measurements.
As used in the description herein and throughout the claims that follow, the meaning of
15 “a,” “an,” and “the” includes plural reference unless the context clearly dictates otherwise.
Also, as used in the description herein, the meaning of “in” includes “in” and “on” unless
the context clearly dictates otherwise.
As used herein, and unless the context dictates otherwise, the term “coupled to” is
20 intended to include both direct coupling (in which two elements that are coupled to each
other contact each other) and indirect coupling (in which at least one additional element
is located between the two elements). Therefore, the terms “coupled to” and “coupled
with” are used synonymously.
25
One component of the present disclosure pertains to a machine for customizing an
assessment version to an evaluation style. The device includes a memory consisting of: a
content material library database containing a plurality of activates and assessment data
related to each of the plurality of prompts; and a version database including a plurality of
30 assessment models for computerized evaluation of acquired person responses. In some
embodiments, the assessment statistics of every of the plurality of activates consists of a
pointer linking to the associated evaluation model. The gadget includes as a minimum
Page 7 of 23
one processor which can: receive a plurality of responses obtained from a plurality of
customers in reaction to presenting of a activate; discover an assessment version
applicable to the supplied spark off, the assessment version along with a machine gaining
knowledge of model educated to output a rating relevant to at least portions of a reaction;
5 generate a education indicator, the training indicator offering a graphical depiction of the
diploma to which the diagnosed evaluation model is skilled; determine a schooling
reputation of the identified version; whilst the model is identified as insufficiently skilled,
acquire as a minimum one assessment input; update education of the evaluation model
primarily based on the at the least one acquired evaluation enter; and manage the training
10 indicator to reflect the diploma to which the assessment model is skilled subsequent to
the updating of the schooling of the assessment version.
In some embodiments, the assessment model includes a plurality of assessment fashions.
In some embodiments, every of the plurality of evaluation fashions is related to an
15 evaluation part of the supplied prompt. In a few embodiments, the as a minimum one
processor can decide a first reaction ordering. In a few embodiments, the primary reaction
ordering identifies an order of offering responses to the person for assessment. In some
embodiments, the response ordering is determined based totally at the anticipated
contribution of each response towards finishing touch of training of the evaluation model.
20
In a few embodiments, the at the least one processor can decide a 2d education repute of
the recognized model next to the updating of the training of the evaluation version based
on the at the least one acquired assessment input. In a few embodiments, the at least one
processor can: car-evaluate the response when the second education status of the
25 diagnosed version is diagnosed as sufficiently trained; decide an acceptability of the autoevaluating of the reaction; and decide a 2nd response ordering when the auto-evaluating
of the reaction is decided as unacceptable.
In a few embodiments, figuring out an evaluation version relevant to the provided set off
30 consists of: figuring out activate evaluation portions; and retrieving a sub-version
associated with each of the recognized spark off assessment portions. In some
embodiments, the at the least one server can determine a schooling degree of the
Page 8 of 23
diagnosed model. In a few embodiments, figuring out a education level of the identified
model includes: retrieving sub-model information for each of the retrieved sub-models;
figuring out a sub-version confidence stage for every of the retrieved sub-models; and
figuring out an combination self assurance level primarily based on a combination of the
5 determined sub-model confidence tiers.
One factor of the prevailing disclosure relates to a method of customizing an assessment
model to an assessment style. The approach includes: receiving a plurality of responses
received from a plurality of users in reaction to imparting of a spark off; identifying an
10 evaluation version relevant to the supplied set off, which assessment model consists of a
device learning model educated to output a rating relevant to at the least portions of a
reaction; generating a schooling indicator, which training indicator offers a graphical
depiction of the diploma to which the identified evaluation version is skilled; figuring out
a training status of the identified model; whilst the model is recognized as insufficiently
15 trained, receiving at the least one evaluation input; updating schooling of the evaluation
model primarily based at the at the least one received evaluation enter; and controlling
the education indicator to mirror the degree to which the evaluation model is skilled
subsequent to the updating of the training of the evaluation model.
20 In some embodiments, the assessment model consists of a plurality of assessment
fashions. In some embodiments, every of the plurality of evaluation models is associated
with an assessment part of the supplied activate. In a few embodiments, the approach
consists of determining a primary response ordering, wherein the primary reaction
ordering identifies an order of supplying responses for evaluation by the user.
25
In some embodiments, the response ordering is determined based totally at the estimated
contribution of each reaction in the direction of final touch of education of the assessment
model. In some embodiments, the method consists of: determining a second training
status of the identified model subsequent to the updating of the training of the evaluation
30 model primarily based on the at the least one obtained assessment input. In some
embodiments, the approach includes: auto-evaluating the reaction when the second
schooling repute of the diagnosed model is identified as sufficiently skilled; determining
Page 9 of 23
an acceptability of the auto-evaluating of the response; and determining a second reaction
ordering when the automobile-comparing of the reaction is decided as unacceptable.
In some embodiments, identifying an evaluation model applicable to the supplied prompt
5 includes: identifying set off assessment portions; and retrieving a sub-version associated
with each of the diagnosed set off assessment portions. In some embodiments, the
technique includes determining a schooling stage of the recognized model. In some
embodiments, figuring out a training degree of the diagnosed model consists of: retrieving
sub-model information for every of the retrieved sub-models; figuring out a sub-model
10 self belief stage for every of the retrieved sub-fashions; and figuring out an combination
self assurance level primarily based on a aggregate of the determined sub-model self
belief tiers.
One issue of the existing disclosure relates to a system for controlling schooling great of
15 a system learning version. The device includes a memory including: a content library
database containing a plurality of prompts and assessment data related to every of the
plurality of prompts; and a version database such as a plurality of evaluation fashions for
automatic assessment of acquired user responses. The system includes at the least one
processor. The at the least one processor can: acquire a plurality of responses acquired
20 from a plurality of customers in reaction to supplying of as a minimum one spark off;
discover an assessment model applicable to the provided set off, which assessment model
consists of a device studying version skilled to output a rating applicable to at the least
portions of a reaction; generate a education indicator, which education indicator presents
a graphical depiction of the degree to which the diagnosed assessment version is trained;
25 decide a training popularity of the diagnosed model; manage the training indicator to
become aware of the schooling popularity of the diagnosed version; automatically
compare the plurality of responses with the assessment model when the model is
recognized as sufficiently trained; generate a graphical indicator of assessment version
performance, which indicator of assessment model performance characterizes an
30 characteristic of the assessment; and control a user interface to show the graphical
indicator of assessment model overall performance.
Page 10 of 23
In a few embodiments, the graphical indicator shows at least one in all: a distribution of
rankings generated via the evaluation version; a confidence level of the assessment
model; and an accuracy degree of the evaluation version. In a few embodiments, the
graphical indicator identifies outlier rankings primarily based on historic user facts. In a
5 few embodiments, identifying outliner ratings based on ancient person records consists
of: retrieving historic statistics; evaluating the historic data to effects of evaluating the
plurality of responses; and indicating an outlier rating while a discrepancy among the
ancient statistics and outcomes of comparing the plurality of responses exceeds a
threshold degree.
10
In a few embodiments, the historic facts consists of a ancient assessment end result
distribution. In a few embodiments, the effects of comparing the plurality of responses
includes an evaluation result distribution. In some embodiments, the historical records
consists of consumer historic statistics. In a few embodiments, the person ancient records
15 pertains to as a minimum one users formerly acquired evaluation results.
In a few embodiments, the at least one processor can decide acceptability of the
assessment. In a few embodiments, the acceptability of the evaluation is decided based
on the diagnosed outlier scores. In some embodiments, the at the least one processor can
20 further train the assessment model when the assessment is unacceptable. In some
embodiments, the at least one processor can: get hold of a selection of at least one
response for reevaluation; receive an assessment enter for the as a minimum one reaction;
and replace schooling of the assessment version based on the obtained evaluation enter
for the at the least one response.
25
One issue of the prevailing disclosure relates to a way of controlling training pleasant of
a machine mastering model. The approach includes: receiving a plurality of responses
obtained from a plurality of customers in response to offering of as a minimum one spark
off; figuring out an assessment model applicable to the supplied prompt, which evaluation
30 version includes a device studying version skilled to output a score applicable to at the
least portions of a response; producing a schooling indicator, which education indicator
presents a graphical depiction of the degree to which the diagnosed assessment model is
Page 11 of 23
educated; figuring out a training reputation of the diagnosed version; controlling the
education indicator to become aware of the education popularity of the diagnosed version;
automatically evaluating the plurality of responses with the assessment version when the
model is recognized as sufficiently trained; producing a graphical indicator of evaluation
5 model performance, which indicator of assessment model performance characterizes an
attribute of the evaluation; and controlling a person interface to display the graphical
indicator of assessment model performance.
In a few embodiments, the graphical indicator indicates at least considered one of: a
10 distribution of rankings generated by way of the assessment model; a self assurance
degree of the evaluation model; and an accuracy level of the evaluation version. In some
embodiments, the graphical indicator identifies outlier scores based on historical
consumer facts. In some embodiments, identifying outliner rankings based on historic
person information consists of: retrieving historic data; evaluating the historic facts to
15 results of comparing the plurality of responses; and indicating an outlier score while a
discrepancy between the historical facts and consequences of comparing the plurality of
responses exceeds a threshold stage.
In a few embodiments, the ancient statistics consists of a ancient assessment end result
20 distribution. In some embodiments, the effects of comparing the plurality of responses
includes an assessment result distribution. In a few embodiments, the historic information
consists of user historic data. In some embodiments, the user historic facts pertains to as
a minimum one customers previously acquired evaluation results.
25 In a few embodiments, the approach consists of determining an acceptability of the
evaluation. In a few embodiments, the acceptability of the assessment is decided primarily
based at the diagnosed outlier scores. In some embodiments, the approach includes
schooling the assessment version whilst the assessment is unacceptable. In a few
embodiments, the method consists of: receiving a diffusion of at least one reaction for
30 reevaluation; receiving an evaluation enter for the as a minimum one reaction; and
updating education of the evaluation model based totally on the received evaluation enter
for the at least one response.
Page 12 of 23
One factor of the prevailing disclosure pertains to a device for schooling a version for a
custom authored prompt. The gadget includes a memory which includes: a content
material library database such as a plurality of prompts; and a version database which
includes a at the least one model educated to assess prompts. The system includes at the
5 least one processor that could: get hold of a spark off; parse the prompt to figuring out a
plurality of prompt evaluation portions associated with the acquired activate; become
aware of pre-present information relevant to one of the assessment portions of the
received spark off; educate a version for evaluating responses to the activate at least in
part based totally at the pre-existing records.
10
In some embodiments, parsing the set off includes identifying a complexity of every of
the plurality of activate evaluation quantities associated with the received spark off. In a
few embodiments, the pre-existing records includes as a minimum considered one of: a
pre-current model skilled to evaluate responses to another set off; and pre-current
15 response facts generated from responses to other activates.
In a few embodiments, the at the least one processor can perceive a writer of the received
set off. In some embodiments, the at least considered one of: the pre-existing version; and
the pre-current reaction information are diagnosed based at the author of the acquired
20 prompt. In a few embodiments, identifying pre-current response statistics includes
identifying response information similar to at the least one of the plurality of activate
evaluation portions through a similarity rating.
In a few embodiments, the as a minimum one processor can iteratively: examine
25 sufficiency of the schooling of the version; and generate new training records when the
schooling of the version is inadequate. In a few embodiments, the processor can offer the
set off to a person; and obtain a response to the furnished spark off. In a few embodiments,
producing new schooling records includes: producing an assessment of the obtained
response; and updating the schooling based totally at the obtained reaction and on the
30 assessment of the received response. In some embodiments, updating the schooling based
at the received response and at the evaluation of the obtained reaction includes: receiving
a plurality of responses and a plurality of critiques to the obtained responses; determining
Description:Page 13 of 23
an ordering to the acquired responses; and schooling based totally at the ordering of the
acquired responses.
In some embodiments, the as a minimum one processor can generate a graphical training
indicator. In some embodiments, the graphical education indicator offers a graphical5
depiction of the diploma to which the identified assessment version is educated. In some
embodiments, the at the least one processor can manipulate the training indicator to mirror
the diploma to which the evaluation model is skilled subsequent to the updating of the
training of the assessment model.
10
One element of the prevailing disclosure relates to a technique of schooling a model for
a custom authored prompt. The method consists of: receiving a set off; parsing the activate
to figuring out a plurality of activate evaluation quantities associated with the received
set off; figuring out pre-current records relevant to one of the assessment portions of the
obtained activate; and automatically training a model for comparing responses to the15
activate at least in component primarily based on the pre-present facts.
In some embodiments, parsing the spark off includes identifying a complexity of each of
the plurality of set off assessment quantities associated with the obtained set off. In some
embodiments, the pre-current information consists of as a minimum considered one of: a20
pre-current model trained to assess responses to any other set off; and pre-existing
response facts generated from responses to different prompts. In some embodiments, the
approach consists of figuring out a creator of the received prompt. In a few embodiments,
the at the least one in every of: the pre-current model; and the pre-current response data
are diagnosed primarily based on the author of the obtained spark off.25
In some embodiments, figuring out pre-current reaction information includes identifying
response statistics similar to at the least one of the plurality of prompt assessment portions
via a similarity rating. In some embodiments, the approach includes iteratively:
comparing sufficiency of the schooling of the model; and producing new training facts30
whilst the schooling of the version is inadequate.
Page 14 of 23
In some embodiments, the technique consists of: providing the set off to a user; and
receiving a response to the supplied activate. In some embodiments, generating new
schooling information includes: generating an evaluation of the received reaction; and
updating the education based totally at the acquired reaction and at the assessment of the
acquired response. In a few embodiments, updating the education based on the acquired5
reaction and on the assessment of the obtained response consists of: receiving a plurality
of responses and a plurality of critiques to the obtained responses; determining an
ordering to the obtained responses; and education primarily based at the ordering of the
acquired responses.
10
In a few embodiments, the approach includes generating a graphical training indicator. In
a few embodiments, the graphical education indicator presents a graphical depiction of
the degree to which the recognized evaluation version is skilled. In some embodiments,
the method includes controlling the training indicator to reflect the diploma to which the
evaluation model is skilled subsequent to the updating of the education of the evaluation15
model.
One thing of the present disclosure relates to a system for schooling a model for a custom
authored activate evaluation. The gadget includes a reminiscence inclusive of: a content
library database consisting of a plurality of prompts; and a version database which include20
a at the least one model educated to evaluate prompts. The machine includes as a
minimum one processor which can: iteratively acquire a set off from a consumer via a set
off creation window within a user interface; offer iterative comments to a person via the
set off advent window; parse the prompt to figuring out a plurality of set off assessment
quantities associated with the obtained activate; discover pre-present records relevant to25
one of the evaluation quantities of the received set off; educate a model for evaluating
responses to the spark off as a minimum in element primarily based at the pre-existing
facts; and provide training information to the person via a training degree indicator within
the consumer interface.
30
In a few embodiments, the at least one processor can iteratively: collect new training data;
update education of the model with the new education information; and compare the
Page 15 of 23
schooling of the version. In a few embodiments, the at the least one processor can
determine that the model is satisfactorily trained. In some embodiments, the at the least
one processor can control an evaluation interface to generate a release window upon
determining that the version is satisfactorily skilled. In a few embodiments, the launch
window affords as a minimum one in every of: an indicator of sufficiency of training of5
the model; and a function manipulable to provoke automobile assessment with the version
of responses received to the activate.
In a few embodiments, the as a minimum one processor can: provide the prompt to a
plurality of customers; acquire a response from every of the plurality of customers;10
determine that the model is satisfactorily skilled; and car-compare the responses with the
version. In a few embodiments, the at least one processor can manipulate the assessment
interface to display as a minimum certainly one of the car-evaluated responses. In some
embodiments, the as a minimum one processor can receive a amendment to the displayed
at the least one of the car-evaluated responses through an input characteristic of the15
evaluation interface.
In some embodiments, the as a minimum one processor can control the evaluation
interface to alter an look of the input feature in reaction to the acquired amendment. In
some embodiments, the at the least one processor can manage generation of an output20
data interface. In some embodiments, the output information interface consists of a
scoring summary window figuring out scoring reputation of received responses. In a few
embodiments, the output records interface includes a distribution window such as a
graphical show of a distribution of the auto-opinions of the responses.
25
One component of the existing disclosure relates to a technique for training a version for
custom authored spark off assessment. The method consists of: iteratively receiving a
spark off from a user via a spark off introduction window within a consumer interface;
imparting iterative comments to a user through the activate advent window; parsing the
spark off to figuring out a plurality of activate evaluation quantities related to the acquired30
prompt; identifying pre-existing information applicable to one of the assessment
quantities of the received spark off; training a version for comparing responses to the set
Page 16 of 23
off as a minimum in part based on the pre-present records; and providing training data to
the user via a training degree indicator within the person interface.
In a few embodiments, the technique includes iteratively: accumulating new schooling
information; updating education of the model with the new schooling information; and5
evaluating the education of the version. In a few embodiments, the approach includes
determining that the model is adequately trained. In some embodiments, the method
consists of controlling an assessment interface to generate a release window upon
determining that the model is sufficiently trained. In some embodiments, the launch
window gives at least one in all: a hallmark of sufficiency of education of the version;10
and a characteristic manipulable to initiate vehicle evaluation with the version of
responses received to the activate.
In a few embodiments, the approach consists of: imparting the prompt to a plurality of
users; receiving a reaction from every of the plurality of customers; determining that the15
version is satisfactorily educated; and vehicle-comparing the responses with the model.
In a few embodiments, the method includes controlling the assessment interface to
display as a minimum one of the car-evaluated responses. In a few embodiments, the
approach consists of receiving a modification to the displayed at least one among the
auto-evaluated responses through an enter characteristic of the evaluation interface.20
In some embodiments, the technique consists of controlling the assessment interface to
regulate an look of the enter function in reaction to the acquired amendment. In some
embodiments, the approach consists of controlling generation of an output data interface.
In a few embodiments, the output information interface consists of a scoring precis25
window figuring out scoring reputation of acquired responses. In some embodiments, the
output facts interface includes a graphical show of the car-critiques of the responses.
One issue of the prevailing disclosure relates to a gadget for interface-based totally
evaluation output customization. The gadget consists of a memory consisting of: a content30
material library database containing a plurality of activates and assessment data
associated with each of the plurality of prompts; and a version database inclusive of a
Page 17 of 23
plurality of evaluation models for automated evaluation of acquired person responses. In
some embodiments, the assessment statistics of each of the plurality of prompts includes
a pointer linking to the related assessment model. The gadget can consist of at the least
one processor which could: get hold of a plurality of responses from a plurality of users
to a supplied prompt; evaluate the received plurality of responses with an assessment5
version, the evaluation version which include a device studying version trained to output
a rating relevant to as a minimum quantities of a reaction; generate evaluation facts
characterizing at the least one characteristic of the evaluated plurality of responses; and
generate an output panel consisting of at least one overall performance change interface,
which overall performance change identifies an characteristic of the evaluated plurality10
of responses, and which overall performance modification interface includes an input
function. In some embodiments, the input characteristic is user manipulable to trade the
characteristic of the evaluated plurality of responses. The at the least one processor can
acquire a enter through the enter feature; modify the characteristic of the evaluated
plurality of responses; and generate updated evaluation information primarily based at15
least in element on the modified attribute of the evaluated plurality of responses.
In some embodiments, the attribute of the evaluated plurality of responses consists of a
rating distribution generated via the assessment version. In a few embodiments, the input
acquired thru the input characteristic adjustments at the least one among: a form of the20
rating distribution; a width of the rating distribution; and a middle of the rating
distribution. In some embodiments, comparing the obtained responses consists of
producing a first score for each of the obtained responses. In a few embodiments,
generating up to date assessment records includes producing a 2d rating for every of the
obtained responses. In some embodiments, the second one score is generated at least in25
component based on the enter received via the input function.
In some embodiments, the output panel further includes a model panel characterizing at
the least one characteristic of the evaluation version. In some embodiments, the as a
minimum one attribute of the evaluation version consists of at least one among: a30
commonplace assessment parameter; and a version identifier. In some embodiments, the
typical assessment parameter identifies: a selected ordinary assessment parameter; and an
Page 18 of 23
application stringency. In a few embodiments, the well-known assessment parameter
consists of as a minimum one among: a formatting style; a proficiency level; and a
language. In a few embodiments, the output panel consists of choice function wherein a
person can select one among a plurality of prevalent evaluation parameters. In some
embodiments, the assessment model education is based at the least in a part of every of5
the plurality of customary evaluation parameters.
One component of the existing disclosure relates to a method for interface-based totally
assessment output customization. The approach consists of: receiving a plurality of
responses from a plurality of users to a supplied spark off; comparing the acquired10
plurality of responses with an evaluation model, the assessment version along with a
device gaining knowledge of model skilled to output a rating relevant to as a minimum
quantities of a response; producing evaluation records characterizing at the least one
characteristic of the evaluated plurality of responses; and generating an output panel
including at the least one performance change interface. In a few embodiments, the15
performance change identifies an characteristic of the evaluated plurality of responses. In
a few embodiments, the performance change interface includes an input characteristic. In
some embodiments, the input characteristic is consumer manipulable to exchange the
characteristic of the evaluated plurality of responses. The method can consist of: receiving
a enter through the enter feature; modifying the characteristic of the evaluated plurality20
of responses; and producing updated evaluation facts based totally as a minimum in
element on the modified characteristic of the evaluated plurality of responses.
In a few embodiments, the characteristic of the evaluated plurality of responses includes
a score distribution generated with the aid of the evaluation model. In some embodiments,25
the input obtained thru the input function modifications at least considered one of: a form
of the score distribution; a width of the rating distribution; and a middle of the score
distribution. In some embodiments, evaluating the received responses consists of
generating a primary score for each of the received reaction; and wherein generating
updated assessment facts consists of generating a 2d score for each of the obtained30
responses. In some embodiments, the second rating is generated at least in component
based at the enter acquired through the input characteristic.
Page 19 of 23
In a few embodiments, the output panel further consists of a version panel characterizing
at the least one characteristic of the evaluation model. In a few embodiments, the at the
least one characteristic of the assessment model includes as a minimum considered one
of: a popular assessment parameter; and a version identifier. In some embodiments, the
ordinary evaluation parameter identifies: a particular widely wide-spread evaluation5
parameter; and an software stringency. In some embodiments, the usual evaluation
parameter consists of at least one among: a formatting style; a talent level; and a language.
In a few embodiments, the output panel consists of choice function wherein a user can
choose one among a plurality of universal assessment parameters. In some embodiments,
the assessment version training is based at the least in part of every of the plurality of10
popular assessment parameters.
Further areas of applicability of the present disclosure becomes apparent from the certain
description provided hereinafter. It ought to be understood that the distinctive description
and particular examples, while indicating diverse embodiments, are meant for purposes15
of instance most effective and aren't supposed to always limit the scope of the disclosure
Unless the context dictates the contrary, all ranges set forth herein should be interpreted
as being inclusive of their endpoints, and open-ended ranges should be interpreted to20
include commercially practical values. Similarly, all lists of values should be considered
as inclusive of intermediate values unless the context indicates the contrary.
Regarding terms used herein, it should also be understood the terms are for the purpose
of describing some particular embodiments, and the terms do not limit the scope of the25
concepts provided herein. Ordinal numbers (e.g., first, second, third, etc.) are generally
used to distinguish or identify different features or steps in a group of features or steps,
and do not supply a serial or numerical limitation. For example, “first,” “second,” and
“third” features or steps need not necessarily appear in that order, and the particular
embodiments including such features or steps need not necessarily be limited to the three30
features or steps. Labels such as “left,” “right,” “front,” “back,” “top,” “bottom,”
“proximal,” “distal,” and the like are used for convenience and are not intended to imply,
Page 20 of 23
for example, any particular fixed location, orientation, or direction. Instead, such labels
are used to reflect, for example, relative location, orientation, or directions. Singular
forms of “a,” “an,” and “the” include plural references unless the context clearly dictates
otherwise.
5
Aspects of the machine and techniques are described beneath with reference to illustrative
embodiments. The references to illustrative embodiments underneath are not made to
restriction the scope of the claimed subject matter. Instead, illustrative embodiments are
used to useful resource inside the description of numerous aspects of the device. The
description, made by means of way of example and reference to illustrative reference is10
not meant to be limiting as regards any component of the claimed problem remember.
It should be understood that even as some of the foregoing strategies include an actor
including a clinician or a person which include employee of a biomedical lab, CSR, or
the like, each method of such strategies consists of at least that clinician or worker. In15
different words, the clinician or employee in such techniques can be a couple of clinician
or worker relying upon one or extra occasions. For instance, the clinician or worker in
such methods can be extraordinary clinicians or employees due to a exchange in shifts,
for instance, a change in an afternoon shift to a swing shift.
20
While a few unique embodiments were disclosed herein, and whilst the particular
embodiments had been disclosed in a few detail, it isn't the aim for the precise
embodiments to restriction the scope of the principles provided herein. Additional
variations and/or changes can appear to the ones of everyday skill inside the artwork, and,
in broader aspects, those adaptations and/or changes are encompassed as properly.25
Accordingly, departures may be made from the specific embodiments disclosed herein
without departing from the scope of the principles provided herein. , C , C , Claims:Page 21 of 23
CLAIMS,
We Claims
[CLAIM 1] An investigation and analysis of machine learning techniques to measure
students’ performance comprising:
A data processing unit for making ready the very last records set of
students, the initial information set consisting of students and the very
statistics set after pre-processing students who were a combination of
college students; a characteristic extraction processing unit to extract the
virtually vital capabilities and put off all redundant and unimportant
capabilities; a type processing unit for classifying the students based
totally on their educational performance using special class algorithms,
namely logistic regression; an evaluation processing unit for acting an
assessment of the instructional overall performance prediction gadget
through calculating the metrics of type accuracy, sensitivity, specificity
and vicinity underneath; and a result interpretation processing unit to gain
a proof in the back of the consequences this is interpretable by means of
human beings, the use of an approach known as to acquire said
explanation.
[CLAIM 2] The investigation and analysis of machine learning techniques to measure
students’ performance as claimed in claim 1, wherein machine learning
system filter the dataset of the student’s performance, which is play in
provided tasks.
[CLAIM 3] The investigation and analysis of machine learning techniques to measure
students’ performance as claimed in claim 1, wherein machine learning
system enable to education institutes to predict the academic performance
of the interacting students.
Page 22 of 23
[CLAIM 4] The investigation and analysis of machine learning techniques to measure
students’ performance as claimed in claim 1, wherein machine learning
system algorithm for Feature Selection, Classification and Clustering
of the students’ performance