Journal of Survey in Fisheries Sciences 10(1S) 2030-2039 2023

A Novel Ranking Approach to Improved

Health Insurance Cost Prediction by
Comparing Linear Regression to Random
Forest
G.Satya Mounika Kalyani ¹, Rama Parvathy L²*
¹Research Scholar, Saveetha School of Engineering, Saveetha Institute of Medical and Technical
Sciences, Saveetha University, Tamil Nadu, India, PinCode: 602105.
²*Project Guide, Corresponding Author, Saveetha School of Engineering, Saveetha Institute of
Medical and Technical Sciences, Saveetha University, Tamil Nadu, India, PinCode: 602105

ABSTRACT
Aim: The aim of research is to classify the cost prediction in health insurance using novel ranking with machine
learning algorithms. Materials and Methods: The categorizing is performed by adopting a sample size of n=10
in Linear Regression and sample size n = 10 in random forest algorithms was iterated 20 times for efficient and
accurate analysis on labeled images with G power in 80% and threshold 0.05%, CI 95% mean and standard
deviation. Results: The analysis of the results shows that the Linear Regression has a high accuracy of
(92.53%) in comparison with the random forest algorithm (91.68 %). There is a statistically significant
difference between the study groups with (p<0.05). Conclusion: Prediction in classifying the cost prediction in
health insurance shows that the Linear Regression appears to generate better accuracy than the cost prediction
random forest algorithm.

Keywords: Health Care, Machine Learning, Random Forest Regression, Novel Ranking, Linear Regression.

INTRODUCTION chance. Ignorance of the problem

The purpose of the research is to
improve the accuracy of cost prediction in
health insurance using novel ranking with
machine learning algorithms (Rajczi
2019). It preimates the sequence of
predicted health cost insurance using linear
regression over random forest. It is found
to be important in today’s world since
health insurance is more convenient for
the patients with respect to various
scenarios (Quadagno 2006) a widespread
statistical phenomenon with potentially
serious implications for health care. It can
result in wrongly concluding that an effect
is due to treatment when it is due to
will lead to errors in decision making
(Muremyi et al. 2020; Baum et al. 2021;
Pauly 2015). Doctors observe the views of
different patients to make decisions. The
applications of predicting health care costs
for each patient with high certainty,
problems such as accountability could be
solved, enabling control over all parties
involved in patients’ care. It could also be
used for other applications such as risk
assessment in the health insurance
business, allowing competitive premium
charges, or for the application of new
policies by governments to improve public
health. (Rajczi 2019; Quadagno 2006)

2030
Rama Parvathy L.et.al.,A Novel Ranking Approach to Improved Health Insurance Cost Prediction by
Comparing Linear Regression to Random Forest
In distinguishing and forecasting
cost prediction in health care using novel
ranking by comparing machine learning
algorithm 880 journals from IEEE Xplore
digital library, 480 articles from
ScienceDirect, 1430 articles from google
scholar and 498 articles from Linear
regression and Random forest algorithm
are two highly correlated parts in
prediction of cost. In this project is to
create a model based on statistically
significant factors independent variables
which will affect premium charges
dependent variables by an insurance
company (Qudsi 2015). Among all the
articles and journals the most cited paper is
(Kharrazi et al. 2021) nearly 1340 people
cited these articles and this article is
useful for future research. An application
is also built which can be interlinked with
the linear regression over random forest
model to view the result on UI based on
the input parameters (Burns 2015). While
the health care law in any country does
have some rules for companies to follow to
determine premiums.concerning the value
of insurance in the lives of individuals, it
becomes important for the companies of
insurance to be sufficiently precise to
measure or quantify the amount covered
by this policy and the insurance charges
which must be paid for it. The model is
trained on insurance data from the past
(Nakamura et al. 2021; Rushing 1986).
The requisite factors to measure the
payments can then be defined as the
model inputs, then the model can
correctly anticipate insurance (Shiny Irene
et al. 2021). The use of multiple linear
regression in this analysis since there are
many independent variables used to
calculate the dependent (target) variable.
For this study, the dataset for cost of health
insurance is used preprocessing of the
datasets (Cebul et al. 2008).
(Bhavikatti et al. 2021; Karobari et
al. 2021; Shanmugam et al. 2021; Sawant
et al. 2021; Muthukrishnan 2021; Preethi
et al. 2021; Karthigadevi et al. 2021;
Bhanu Teja et al. 2021; Veerasimman et
al. 2021; Baskar et al. 2021)
The methods which were used
before have less accuracy and detection
rate in predicting cost in health insurance.
Determine premiums,it’s really up to know
the most important factors and how much
statistical importance they hold. In order to
sequence the methods and techniques in
this research study Linear Regression
generally fares better than Random Forest
(Rajczi 2019). It also takes more time to
train a Random forest approach for
analyzing health Insurance datasets which
are based on the model to increase with the
size of the datasets. Its estimates and
calculations are probably done using a
characteristic technique. Applying Linear
regression.over Random forest model to
Medical Insurance dataset to predict future
insurance costs for the individuals
(Willemse-Duijmelinck et al. 2017).
Machine learning is a method of data
analysis which sends instructions
(programmable code) to code to computers
so that they can learn from data (Cabrera
2011). Then, based on the learned
data,they provide us with the predicted
results/patterns. The aim of the research
work is to predict the cost in health
insurance using machine learning and
comparing ML techniques Linear
Regression over Random Forest.
MATERIALS AND METHODS
The research work was performed in
the Image Processing Lab, Department of
2031
Journal of Survey in Fisheries Sciences 10(1S) 2030-2039 2023

Computer Science and Engineering, there is one dependent variable and

Saveetha School of Engineering, Saveetha multiple independent variables. The value
Institute of Medical and Technical of a dependent variable is miscalculated
Sciences, Chennai. Basically it is from independent variables. The
considered that two groups of classifiers dependent variable is medical charges and
are used, namely Linear Regression and independent variables are
Random Forest algorithms, which are used age,gender,smoker,BMI,children,region.
to predict cost in health insurance. Group 1 Multiple Linear Regression uses the
is the Linear Regression with the sample ordinary least squares method to find a
size of 10 and the Random Forest is Group best fitting line,which involves multiple
2 with sample size of 10 and they are independent variables.
compared for more Accuracy score and
Precision score values for choosing the Pseudocode for Linear Regression
best algorithm. The Pre- test analysis has Import Linear Regression as LR
been prepared by having a G power of Import pandas as pd
80% and threshold 0.05%, CI 95% mean ImportMatplotlib.pyplot as plt
and standard deviation (Muremyi et al. Compare from sklearn ensemble import
2020). Sample size has been calculated Linearregression
and it is identified that 10 samples/ group From sklearn .tree import
in total 20 samples with a standard RandomForestClassifier
deviation for Linear Regression = 95.2% Data extraction from sklearn
and Random Forest = 93.3%.Sample size Initiate sklearn.metrics import accuracy
has been calculated and it is identified as price
standard deviation for Linear regression = Calculate sequence
0.20359 and Random forest = 0.24832. sklearn.mode_selection import
A minimum of 4GB ram is required train_test_split
to use and install all the necessary Find results from
applications, and a minimum of i3 sklearn.feature_extraction.value import
processor to run all the applications and CountVectorizer
processes simultaneously. A minimum of count_vectorizer=CountVectorizer()
50GB hard disk space is required to install Cv.Shape
the required software and files, and an lrg=LRclassifier(max_depth=8,estimators
internet connection is required to =100,nthread=3)
download and install all the necessary lrgc.fit(x_train,y_train)
software and development environment to prediction_lrg=lrgc.predict(x_test)
run this Novel Effective framework. print(accuracy_price(prediction_lrg,y_test)
Python programming language is used for
the application. The version of python Random Forest
used is 3.9, and the IDLE is used to run Random forest is a Machine Learning
and execute the application. algorithm which is used for classification
problems, it is a predictive analysis
Linear Regression algorithm and based on the concept of
Multiple Linear Regression is a probability it will make an exact sequential
basic, machine regression model,in which

2032
Rama Parvathy L.et.al.,A Novel Ranking Approach to Improved Health Insurance Cost Prediction by
Comparing Linear Regression to Random Forest

identification using characteristic Sample T-test was performed.The

technique as shown below. independent data sets are groups,accuracy
and details. The Dependent values are
Pseudocode for Random Forest patient data and cost. A detailed analysis
Import random forest Classifier has been done on these values for finding
Import random forest as rm health insurance in machine learning.
Compare from sklearn.ensemble import
RandomForestClassifier RESULTS
Skip from sklearn.linear_model import The dataset is provided which selects
random forest the random samples from a given dataset
Data extraction from sklearn import svm for health insurance. The data is collected
Initiate sklearn.metrics import accuracy for a period of 10 days.Group Statistics of
score linear regression with random forest by
grouping the iterations with sample size
Calculate sequence from 10,Mean = 0.95.Group Statistics of Linear
sklearn.model_selection import Regression with Random Forest by
train_test_split grouping the iterations with Sample size
Find results from 10, Mean = 95.2000 , Standard Deviation
sklearn.feature_extraction.text import =1.50259 , Standard Error Mean =
CountVectorizer 0.47516. The data is collected for a period
count_vectorizer of 10 days.The model efficiently to
=CountVectorizer(stop_words='english') analyze health insurance.The mean
cv = difference, standard deviation difference
count_vectorizer.fit_transform(data['Clean and significant difference of Linear
_TweetText']) Regression based cost prediction and
cv.shape Random Forest based cost prediction is
Rm= Random forest() tabulated in Table 3,which shows there is a
Rm.fit(X_train,y_train) significant difference between the two
prediction_Rm = Rm.predict(X_test) groups since p<0.001 with an independent
print(accuracy_score(predi Rm sample T-Test. Table 1Health Insurance
action_rm,y_test)) data set with five attributes which selects
the random samples from a given dataset.
Statistical Analysis Table 2 the dataset is the independent and
The analysis was done using IBM dependent variable.Fig. 1 represents the
SPSS version 21. It is a statistical software comparison of Linear Regression over
tool used for data analysis. For both Random Forest in terms of mean
proposed and existing algorithms 10 accuracy. The dependent variables in
iterations were done with a maximum of Health Insurance cost prediction are
15 samples and for each iteration the predicted with the help of the independent
predicted accuracy was noted for variables. The statistical analysis of two
analyzing accuracy. There is a statistically independent groups shows that the Linear
significant difference between the study Regression has a higher accuracy mean
groups with (p<0.05). The value obtained (95.2%) compared to Random Forest.
from the iterations of the Independent

2033
Journal of Survey in Fisheries Sciences
DISCUSSION
In this research clustering has
proven to be a highly effective and
versatile approach for prediction (Kharrazi
et al. 2021; Muremyi et al. 2020). The
applied unsupervised learning with R
explains clustering methods,distribution
analysis,data encoders,and features of R
that enable us to to understand the given
data better and get answers to our point of
prediction problems. The most important
features of detecting cost in health
insurance using Linear Regression is
empirically proven to be a highly effective
than Random Forest classifier.It was
observed that the important factors and the
concept of health insurance(Muremyi et al.
2020; Baum et al. 2021)).
It was observed that the important
factors and the concept of cost prediction
has been analyzed based on the matrix for
health insurance (Willemse-Duijmelinck et
al. 2017). The discussion offers insights
into the reasons for the development of the
practical approach, a concrete
methodology for its implementation and
strategic and tactical applications of the
cost prediction (Maharani et al. 2019). In
this the linear regression is also used for
predicting the output of the given data.
Thus the linear regression shows the more
accurate results in it (Daniel 2015) various
factors which affect the predicted amount
were examined. It was observed that a
person's age and smoking status affects the
prediction most in every algorithm
applied. Attributes which had no effect on
the prediction were removed from the
features. The effect of various independent
variables on the premium amount was also
checked. The attributes also in
combination were checked for better
accuracy results. Premium amount
10(1S) 2030-2039 2023
prediction focuses on a person's own
health rather than other companies'
insurance terms and conditions. The
models can be applied to the data collected
in coming years to predict the premium.
This can help not only people but also
insurance companies to work in tandem
for better and more health centric
insurance amounts (Rice and Thorpe
1993).
Hence the study results produce
clarity in performance with both
experimental and statistical analysis, but it
has some limitations to the proposed work
such as threshold and precision. The
accuracy level of cost prediction in health
care using novel based techniques can still
be improved by implementing artificial
intelligence techniques to predict and
analyze results while comparing with
existing machine learning algorithms. In
the future, the large dataset for health
insurance can be considered to validate
our proposed model with respect to recent
scenarios (Daniel 2015).
CONCLUSION
The current study focused on
machine learning algorithms, Linear
Regression over Random Forest for
higher classification in cost prediction. It
can be slightly improved based on the
Random Forest data sets analysis in
future. The outcome of the study shows
Linear Regression 95.2% higher accuracy
than Random Forest 93.3%.
DECLARATIONS
Conflict of Interests
No conflict of interests in this manuscript
Authors Contribution
Author GSMK was involved in
data collection, data analysis, manuscript
2034
Rama Parvathy L.et.al.,A Novel Ranking Approach to Improved Health Insurance Cost Prediction by
Comparing Linear Regression to Random Forest

writing. Author DV was involved in the Ruby Mishra, S. Sivasaravanan, and D.

Action process, Data verification and Thanikaivel Murugan. 2021. “Detailed
validation, and Critical review of Analysis on Sterculia Foetida Kernel
manuscript. Oil as Renewable Fuel in Compression
Acknowledgement Ignition Engine.” Biomass Conversion
The authors would like to express and Biorefinery, February.
their gratitude towards Saveetha School of https://doi.org/10.1007/s13399-021-
Engineering, Saveetha Institute of Medical 01328-w.
and Technical Sciences ( Formerly known 4. Bhavikatti, Shaeesta Khaleelahmed,
as Saveetha University) for providing the Mohmed Isaqali Karobari, Siti Lailatul
necessary infrastructure to carry out this Akmar Zainuddin, Anand Marya,
research work successfully. Sameer J. Nadaf, Vijay J. Sawant,
Funding Sandeep B. Patil, Adith Venugopal,
We thank the following Pietro Messina, and Giuseppe
organization for providing financial Alessandro Scardina. 2021.
support that enabled us to complete the “Investigating the Antioxidant and
study. Cytocompatibility of Mimusops Elengi
1. SNEW.AI Technologies, Chennai. Linn Extract over Human Gingival
2. Saveetha University Fibroblast Cells.” International
3. Saveetha Institute of Medical and Journal of Environmental Research
Technical Sciences and Public Health 18 (13).
4. Saveetha School of Engineering https://doi.org/10.3390/ijerph18137162
.
REFERENCES 5. Burns, Sarah K. 2015. “Health Care
1. Baskar, M., R. Renuka Devi, J. Analysis for the MCRMC Insurance
Ramkumar, P. Kalyanasundaram, M. Cost Model.”
Suchithra, and B. Amutha. 2021. https://doi.org/10.21236/ada618075.
“Region Centric Minutiae Propagation 6. Cabrera, Delia. 2011. “A Review of
Measure Orient Forgery Detection Algorithms for Segmentation of
with Finger Print Analysis in Health Retinal Image Data Using Optical
Care Systems.” Neural Processing Coherence Tomography.” Image
Letters, January. Segmentation.
https://doi.org/10.1007/s11063-020- https://doi.org/10.5772/15833.
10407-4. 7. Cebul, Randall, James Rebitzer,
2. Baum, Griffin R., Geoffrey Stricsek, Lowell Taylor, and Mark Votruba.
Mathu A. Kumarasamy, Vineeth 2008. “Unhealthy Insurance Markets:
Thirunavu, Gregory J. Esper, Scott D. Search Frictions and the Cost and
Boden, and Daniel Refai. 2021. Quality of Health Insurance.”
“Current Procedural Terminology- https://doi.org/10.3386/w14455.
Based Procedure Categorization 8. Daniel, Ines. 2015. “MARKET
Enhances Cost Prediction of Medicare SEGMENTATION USING COLOR
Severity Diagnosis Related Group in INFORMATION OF IMAGES.”
Spine Surgery.” Spine 46 (6): 391–400. International Journal of Electronic
3. Bhanu Teja, N., Yuvarajan Devarajan, Commerce Studies.

2035
Journal of Survey in Fisheries Sciences 10(1S) 2030-2039 2023

https://doi.org/10.7903/ijecs.1400.
9. Karobari, Mohmed Isaqali, Syed Nahid
Basheer, Fazlur Rahman Sayed,
Sufiyan Shaikh, Muhammad Atif
Saleem Agwan, Anand Marya, Pietro
Messina, and Giuseppe Alessandro
Scardina. 2021. “An In Vitro
Stereomicroscopic Evaluation of
Bioactivity between Neo MTA Plus,
Pro Root MTA, BIODENTINE &
Glass Ionomer Cement Using Dye
Penetration Method.” Materials 14
(12).
https://doi.org/10.3390/ma14123159.
10. Karthigadevi, Guruviah,
Sivasubramanian Manikandan,
Natchimuthu Karmegam, Ramasamy
Subbaiya, Sivasankaran
Chozhavendhan, Balasubramani
Ravindran, Soon Woong Chang, and
Mukesh Kumar Awasthi. 2021.
“Chemico-Nanotreatment Methods for
the Removal of Persistent Organic
Pollutants and Xenobiotics in Water -
A Review.” Bioresource Technology
324 (March): 124678.
11. Kharrazi, Hadi, Xiaomeng Ma, Hsien-
Yen Chang, Thomas M. Richards, and
Changmi Jung. 2021. “Comparing the
Predictive Effects of Patient
Medication Adherence Indices in
Electronic Health Record and Claims-
Based Risk Stratification Models.”
Population Health Management,
February.
https://doi.org/10.1089/pop.2020.0306.
12. Maharani, Dian, The authors are with
Universitas Indonesia, Indonesia,
Hendri Murfi, and Yudi Satria. 2019.
“Performance of Deep Neural Network
for Tabular Data — A Case Study of
Loss Cost Prediction in Fire
Insurance.” International Journal of
Machine Learning and Computing.
https://doi.org/10.18178/ijmlc.2019.9.6
.866.
13. Muremyi, Roger, Dominique
Haughton, Ignace Kabano, and
François Niragire. 2020. “Prediction of
out-of-Pocket Health Expenditures in
Rwanda Using Machine Learning
Techniques.” The Pan African Medical
Journal 37 (December): 357.
14. Muthukrishnan, Lakshmipathy. 2021.
“Nanotechnology for Cleaner Leather
Production: A Review.”
Environmental Chemistry Letters 19
(3): 2527–49.
15. Nakamura, Haruyo, Floriano Amimo,
Siyan Yi, Sovannary Tuot, Tomoya
Yoshida, Makoto Tobe, Md Mizanur
Rahman, Daisuke Yoneoka, Aya
Ishizuka, and Shuhei Nomura. 2021.
“Developing and Validating
Regression Models for Predicting
Household Consumption to Introduce
an Equitable and Sustainable Health
Insurance System in Cambodia.” The
International Journal of Health
Planning and Management, July.
https://doi.org/10.1002/hpm.3269.
16. Pauly, Mark. 2015. “Cost-
Effectiveness Analysis and Insurance
Coverage: Solving a Puzzle.” Health
Economics.
https://doi.org/10.1002/hec.3044.
17. Preethi, K. Auxzilia, K. Auxzilia
Preethi, Ganesh Lakshmanan, and
Durairaj Sekar. 2021. “Antagomir
Technology in the Treatment of
Different Types of Cancer.”
Epigenomics.
https://doi.org/10.2217/epi-2020-0439.
18. Quadagno, Jill. 2006. “Cost
Containment versus National Health
Insurance.” One Nation,
UninsuredWhy the U.S. Has No
National Health Insurance.

2036
Rama Parvathy L.et.al.,A Novel Ranking Approach to Improved Health Insurance Cost Prediction by
Comparing Linear Regression to Random Forest
https://doi.org/10.1093/acprof:oso/978
0195160390.003.0006.
19. Qudsi, Dini Hidayatul. 2015.
Predictive Data Mining of Chronic
Diseases Using Decision Tree: A Case
Study of Health Insurance Company in
Indonesia.
20. Rajczi, Alex. 2019. “Personal Cost.”
The Ethics of Universal Health
Insurance.
https://doi.org/10.1093/oso/978019094
6838.003.0004.
21. Rice, Thomas, and Kenneth E. Thorpe.
1993. “Income-Related Cost Sharing in
Health Insurance.” Health Affairs.
https://doi.org/10.1377/hlthaff.12.1.21.
22. Rushing, William A. 1986. “Health
Insurance, Cost Containment, and
Social Conflict: A Future Perspective.”
Social Functions and Economic
Aspects of Health Insurance.
https://doi.org/10.1007/978-94-009-
4231-8_9.
23. Sawant, Kashmira, Ajinkya M. Pawar,
Kulvinder Singh Banga, Ricardo
Machado, Mohmed Isaqali Karobari,
Anand Marya, Pietro Messina, and
Giuseppe Alessandro Scardina. 2021.
“Dentinal Microcracks after Root
Canal Instrumentation Using
Instruments Manufactured with
Different NiTi Alloys and the SAF
System: A Systematic Review.” NATO
Advanced Science Institutes Series E:
Applied Sciences 11 (11): 4984.
24. Shanmugam, Vigneshwaran, Rhoda
Afriyie Mensah, Michael Försth,
TABLES AND FIGURES
Table 1. Health Insurance datasets with seven attributes which selects the random samples
from a given dataset. Implement Linear Regression and Random Forest for each data point.
Age Sex Bmi Children Smoker
19 FEMALE 27.9 0 YES
Gabriel Sas, Ágoston Restás, Cyrus
Addy, Qiang Xu, et al. 2021. “Circular
Economy in Biocomposite
Development: State-of-the-Art,
Challenges and Emerging Trends.”
Composites Part C: Open Access 5
(July): 100138.
25. Shiny Irene, D., V. Surya, D. Kavitha,
R. Shankar, and S. John Justin
Thangaraj. 2021. “An Intellectual
Methodology for Secure Health
Record Mining and Risk Forecasting
Using Clustering and Graph-Based
Classification.” Journal of Circuits
Systems and Computers 30 (08):
2150135.
26. Veerasimman, Arumugaprabu,
Vigneshwaran Shanmugam,
Sundarakannan Rajendran, Deepak
Joel Johnson, Ajith Subbiah, John
Koilpichai, and Uthayakumar
Marimuthu. 2021. “Thermal Properties
of Natural Fiber Sisal Based Hybrid
Composites – A Brief Review.”
Journal of Natural Fibers, January, 1–
11.
27. Willemse-Duijmelinck, Daniëlle M. I.
D., Daniëlle M. I. Willemse-
Duijmelinck, Wynand P. M. M. van de
Ven, and Ilaria Mosca. 2017.
“Supplementary Insurance as a
Switching Cost for Basic Health
Insurance: Empirical Results from the
Netherlands.” Health Policy.
https://doi.org/10.1016/j.healthpol.201
7.08.003.
Region Charges
SOUTHWEST 16884.924
2037
Journal of Survey in Fisheries Sciences 10(1S) 2030-2039 2023

18 MALE 33.77 1 NO SOUTHEAST 1725.5523

28 MALE 33 3 NO SOUTHEAST 4449.462

33 MALE 22.705 0 NO NORTHWEST 21984.471

32 MALE 28.88 0 NO NORTHWEST 3866.8552

Table 2. Group Statistics of Linear Regression with Random Forest by grouping the
iterations with Sample size 10, Mean = 95.02000 , Standard Deviation = 1.50259 , Standard
Error Mean = 0.47516. Descriptive Independent Sample Test of Accuracy and Precision is
applied for the dataset in SPSS. Here it specifies Equal variances with and without assuming
a T-Test Score of two groups with each sample size of 10.
Group N Mean Std.Deviatio Std.Error
n Mean

Accuracy LR 10 95.2000 1.50259 0.47516

RF 10 93.3000 2.49132 0.78782

Precision LR 10 1.2640 0.20359 0.6438

RF 10 2.4620 0.24832 0.7853

Table 3. Independent Sample Test of Accuracy and Precision ( Calculate P-value <0.001 and
Significant value= 0.107, Mean Difference= 1.9 and confidence interval = (0.92- 0.10).
Logistic Regression and Random Forest are significantly different from each other.
F Sig T Df Sig(2. Mean Std error lower upper
tailed differe difference
nce

Accu Equal 2.885 0.107 2.065 18 <.001 1.9 0.92 -0.03 3.83
racy Variance
Assume

Equal 2.065 14.78 <.001 1.9 0.92 -0.06 3.86

variance
not
assumed

Preci Equal 0.289 0.598 -11.798 18 <.001 -1.198 0.10 -1.41 -0.98
sion Variance
Assume
d

Equal -11.798 17.3 <.001 -1.198 0.10 -1.41 -0.98

variance
not
assumed

2038
Rama Parvathy L.et.al.,A Novel Ranking Approach to Improved Health Insurance Cost Prediction by
Comparing Linear Regression to Random Forest

Fig. 1. Comparison of Linear Regression over Random Forest in terms of mean accuracy.It
explores that the mean accuracy is slightly better than random forest and the standard
deviation is moderately improved compared to random forest.Graphical representation of the
bar graph is plotted using group id as X-axis Linear Regression vs Random Forest,Y-Axis
displaying the error bars with a mean accuracy of detection +/-1 SD.

2039