Abhishek Jaiswar

31031523011

Research Paper

Topic :
To predict calories burned during Various physical activities
like walking, running, cycling, & swimming

Submitted By
Abhishek Jaiswar

Roll No: 31031523011

MSc CS – Part II

Department Of Computer Science

Somaiya Vidyavihar University
SK Somaiya college
 Abhishek Jaiswar
31031523011

To predict calories burned during Various physical activities like

walking, running, cycling, & swimming

1. ABSTRACT
This paper is about making a better model to predict calories burned during physical
activities like walking, running, cycling, & swimming. We looked at a lot of data with
different factors such as Heart rate ,Body temperature, activity duration, and how
hard the activity was. Our approach uses machine learning to guess calorie use
accurately. With that Mifflin St. Jeor Equation for calculating Calories Burnt during
various Physical activity This model could help with personal fitness tracking,
managing weight, & improving sports performance.
For this purpose, I’m using Kaggle dataset that contains important elements like
heart rate, body temperature, & activity duration
2. INTRODUCTION
Calories burned is super important in health & fitness. Knowing how many calories
you burn while exercising can really help with weight management, custom training
plans, & overall health. Older ways to estimate calories—like heart rate monitors or
metabolic carts—can be tough to use & expensive. So, there is a big need for tools
that help predict how many calories are used easily.
By developing an accurate calorie burn prediction model, we hope to help people
make better choices regarding their physical activity and enhance their overall well-
being
3. THE KEY GOAL OF THIS STUDY

This study’s main goal is to build a strong & accurate model for predicting calories
burned. This model will help estimate calorie use while people do different physical
activities, like walking, running, cycling, or swimming. We plan to:

• Collect a diverse dataset: We want to gather lots of information. This includes many
individual traits (like age, gender, weight, height), environmental factors (like
temperature & humidity), and activity details (like how long they exercised, the
intensity, and heart rate).
 Abhishek Jaiswar
31031523011

• Feature engineering: It’s important to pull out useful features from our data. Things
like activity type, duration, distance, pace, & heart rate variability will be included.

• Model selection & training: We’ll try out different machine learning methods (like
linear regression, decision trees, random forests) to see which one best helps with
calorie prediction.

• Model evaluation: The chosen model will be checked for performance using
suitable measures. We’ll look at things like mean squared error (MSE), root mean
squared error (RMSE), & mean absolute error (MAE).

• Model deployment: We’ll create a friendly application or API so people can input
their exercise data and get reliable calorie estimates.

4. LITERATURE Review
Firstly Comparing various Regression Models for Calculating Calories Burnt
Prediction.

1.Linear Regression
Definition: A statistical method that models the relationship between a dependent
variable (calories burnt) and one or more independent variables (e.g., exercise
duration, intensity, weight) as a linear equation.
How it works: Linear regression finds the best-fitting line through the data points to
predict the dependent variable based on the independent variables.

Accuracy for Linear Regression is 90.62 % , With Mean Absolute Error 8.39.

Support Vector Regressor (SVR)

Definition: A machine learning algorithm that finds a hyperplane that separates the
data points as closely as possible while maximizing the margin between the
hyperplane and the nearest data points.
How it works: SVR uses kernels to map the data into a higher-dimensional space,
where it becomes easier to find a linear separation. The hyperplane in this higher-
dimensional space corresponds to a non-linear decision boundary in the original
space

Accuracy for Support Vector Regressor is 88.11 % , With Mean Absolute Error 10.62.
Random Forest
 Abhishek Jaiswar
31031523011

Definition: An ensemble learning method that creates a forest of decision trees, each
trained on a random subset of the data and features.
How it works: Random Forest combines the predictions of all the decision trees to
make a final prediction.
This helps to reduce overfitting and improve generalization performance

Accuracy for random forest is 98.09 with Mean absolute error 1.71 .

Decision Tree Regressor

Definition: A machine learning algorithm that creates a tree-like model to predict a
continuous outcome (calories burnt).
How it works: The decision tree splits the data into subsets based on the values of
features, creating branches and leaves.
The final prediction is made based on the leaf node that the input data point falls
into

So the Accuracy for Decision Tree is 96.23% With Mean Absolute Error 3.37 calories

XGBoost Regressor
Definition: An ensemble learning method that uses gradient boosting to create a
strong predictive model by combining multiple weak models (decision trees).
How it works: XGBoost iteratively adds decision trees to the ensemble, each focusing
on correcting the errors of the previous ones. It also employs regularization
techniques to prevent overfitting.
Accuracy for XGBoost Regressor is 98.43 % with lowest error 1.48 calories
 Abhishek Jaiswar
31031523011

5. methodology
5.1 information collection • wearable twist Information: purchase information from
wearable engineering (e. g smartwatches physical fitness trackers) that proctor eye order
GPS Information accelerometer metrics and different appropriate indicators • Operator-
reported Information: compose Information relating to person traits (age sex angle height)
the case of action undertaken length felt drive and environmental factors
5.2 information pre-processing• Information cleaning: pass outliers take inch lost values and
amend inconsistencies inside the Information set • Characteristic engineering: elicit
important Characteristics from the green information including: o time-based
Characteristics: length rate cadence o heart rate-based Characteristics: mean eye order eye
order variability GPS based Characteristics: space top shift Landscape type o accelerometer-
based Characteristics: measure number strength level
5.3 Check Remove and Teaching • exploratory information psychoanalysis (EDA): see the
information to reveal Layouts trends and correlations betwixt Characteristics and kilocalorie
expenditure • Characteristic selection: important the about pertinent Characteristics
exploitation techniques such as arsenic correlation coefficient psychoanalysis and evaluating
have importance • Representation Teaching: school different car acquisition Representations
including: o linear regression: amp square Check to plant amp baseline o decision trees: amp
Check that captures compound relationships done non-linear Methodes o random forests:
associate in nursing tout ensemble facility that Improves truth away combine aggregate
conclusion trees o Nerve-related Webs: amp iron Check adequate of acquisition Complicated
Layouts from comprehensive Information sets

5.4 Check evaluation • evaluation metrics: employ fit metrics care MSE ,RMSE and MAE to
beat the Representations' Effectiveness • hyperparameter tuning: down the
hyperparameters of the elect Check to raise its effectiveness
5.5 Check • web Use: arise associate in nursing visceral Web diligence that enables
Operators to stimulus their action information and get right kilocalorie consumption
estimates • mobile app: make amp versatile diligence that get tie with wearable devices to
render real-time trailing of kilocalorie expenditure • API: plan associate in nursing API to
incorporate the Check with different physical fitness and health Uses by applying this
methodological analysis the aim is to plant amp extremely right and good kilocalorie
anticipation Check empowering individuals to get knowledgeable choices vis-a-vis their
health and physical fitness goals .
 Abhishek Jaiswar
31031523011

Methodology Using
Design a Calories Calculator Using Mifflin St. Jeor Equation to estimate the number of
calories burnt during activities like Walking, Running, Cycling, Swimming.
Parameters for Walking, Running, Cycling: slow, moderate, fast, very fast, hiking.
Swimming: moderate, laps, vigorous
Based on parameters like Duration, Body Weight.
User will be selecting which type of activity they performed and calculate the amount of
calories they burnt.
The Mifflin-St Jeor equation is a widely used tool to determine the resting metabolic rate
[RMR], which is defined as the number of calories burned while the body is in complete rest.
RMR is also known as resting energy expenditure [REE]. The equation was developed by MD
Mifflin and ST St Jeor

Formula
Females: (10*weight [kg]) + (6.25*height [cm]) – (5*age [years]) – 161
Males: (10*weight [kg]) + (6.25*height [cm]) – (5*age [years]) + 5

Multiply by scale factor for activity level:

Sedentary *1.2
Lightly active *1.375
Moderately active *1.55
Active *1.725
Very active *1.9
 Abhishek Jaiswar
31031523011

Limitations of a Calories Calculator Using the Mifflin St. Jeor equating spell the Mifflin St.
Jeor equation is a widely used method to estimate basal metabolic rate (BMR) a important
factor in calorie expenditure it's important to recognize its limitations when used in a calorie
calculator for activities like walking running cycling and swimming.
important Limitations: Individual Variation: Metabolic Rate: Individuals have unique
metabolic rates influenced by genetics hormones and other factors. a stock equation charge
Mifflin second tonne. Jeor may not accurately capture this variability.
Body Composition: Body composition specifically the ratio of muscle to fat extremely
importantly impacts calorie burn during exercise. the equation doesn't level account for this
activity force and duration: strength level: accurately estimating the force of be (light run
vigorous) beat be set. Subjective measures like perceived exertion may not be precise.
Duration Tracking: Reliable tracking of exercise duration notably for activities like walking
and cycling can be difficult notably without specialized equipment.
Environmental Factors: Temperature and Humidity: These factors influence the body's
thermoregulation and can impact calorie expenditure. the equation doesn't read these
variables altitude: higher altitudes beat clear metabolic rate good the equation doesn't
account for this
dietary factors: thermal effect of food (tef): the post tired to digest lot and be food beat
shift. While the equation doesn't directly measure TEF it's a factor that can influence overall
calorie expenditure.
Addressing Limitations: To Improve the Precision of a calorie calculator consider
incorporating the following: Operator Input: Activity Intensity: Allow Operators to specify the
intensity level (light moderate vigorous) using a standardized scale like the Rating of
Perceived Exertion (RPE). Heart Rate Watching: Combine with heart rate Watches to provide
more accurate intensity estimates.
GPS Tracking: Use GPS devices to track distance and duration notably for outdoor activities
like running and cycling. Advanced Procedures:
Metabolic Equivalent of Task (MET): Employ MET values which represent the energy cost of
specific activities relative to rest to estimate calorie expenditure more precisely. Calculator
learning:
Utilize Calculator learning techniques to Examine individual Information (e.g. eye order
action layouts) and down kilocalorie estimates across time personalized recommendations:
personal kilocalorie goals: take factors care mature gender angle and action point to render
personal kilocalorie goals dietary guidance:
bid dietetic advice to back angle direction and general health by addressing these limitations
and incorporating advance characteristics amp kilocalorie Calculator get render further right
and true estimates of kilocalorie consumption empowering Operators to make informed
decisions about their fitness and nutrition goals.
 Abhishek Jaiswar
31031523011

Conclusion
research successfully Combines the Mifflin St. Jeor equating into amp broader prophetic
frame for estimating calories tempered during activities care walk run cycling and swim.
Here's a summary of the important Understandings and conclusions:
Effective Baseline Prediction Representation: The Mifflin St. Jeor equating serves arsenic
amp foundational drive to gauge resting metabolous rates and provides amp ascendable
multiplier factor for action levels.
This equation when supplemented with activity-specific parameters delivers reliable
baseline calorie estimates.
Improved Precision with Calculator learning: Your research demonstrates that advanced
Calculator learning Procedures like XGBoost and Random Forest very importantly
outperform traditional linear methods achieving accuracies of 98.43% and 98.09%
respectively. these methods work person variations and get compound relationships betwixt
variables such as arsenic eye order action strength and duration Characteristic Tech and
Information set Understandings: incorporating variables care eye order variance
environmental factors and activity-specific metrics (e. g rate duration) Improves anticipation
truth. This aligns with modern fitness tracking methods and allows for a more personalized
approach.
Limitations of the Mifflin St. Jeor approach: despite its substitute the equating does not bill
for person metabolous variations trunk paper (muscle-to-fat ratio) or environmental factors
care height and temperature which get determine kilocalorie fire rates proposed
Improvements: to palliate these limitations combine the Mifflin St. Jeor approach with heart
rate Watches, GPS tracking and MET (Metabolic Equivalent of Task) values is recommended.
in addition leverage car acquisition allows for active personalization account for Operator-
specific information trends across time practical Uses: the world of Operator-friendly tools
care Web Uses versatile apps and genus Apis ensures the availability of your Check.
Integration with wearable fitness devices for real-time tracking further Improves usability.
Final Remarks: Your research bridges traditional methods with cutting-edge Calculator
learning to refine calorie estimation during physical activities. away addressing intrinsic
limitations and leverage contemporary engineering it offers amp pragmatic ascendable root
for individuals aiming to care physical fitness and health goals in effect
 Abhishek Jaiswar
31031523011

References
[1] D. Bubnis, “Calculating how many calories are burned in a day,” Medical News Today, 01 January
2020. [Online]. Available: https:// www.medicalnewstoday.com/articles/319731.

[2] K. S. University, “Burning more calories is easier when working out with someone you perceive as
better,” 26 November 2012. [Online]. Available:
https://www.sciencedaily.com/releases/2012/11/121126130938.htm.

[3] B. K. Tingley, “The New Science on How

We Burn Calories,” The New Work Times Magazine, 14 September 2021. [Online]. Available:
https://www.nytimes.com/2021/09/14/ magazine/calories-weight-age.html.

[4] G. Vijayalakshmi and T. Sridurga, “COMPARING MACHINE LEARNING ALGORITHMS FOR

PREDICTING CALORIES BURNED,” JOURNAL OF EMERGING TECHNOLOGIES AND INNOVATIVE
RESEARCH

(JETIR), vol. 10, no. 3, pp. 519-527, March 2023.

[5] S. P. Vinoy and B. Joseph, “Calorie Burn Prediction Analysis Using XGBoost Regressor and Linear
Regression Algorithms,” in Proceedings of the National Conference on Emerging Computer
Applications (NCECA), Kottayam, 2022.

[6] S. S. Ratnakar and V. S, “Calorie Burnt Prediction using Machine Learning,” International Advanced
Research Journal in Science, Engineering and Technology, vol. 9, no. 6, pp. 781-787, June 2022.

[7] R. K. Singh and V. Gupta, “Calories Burnt Prediction using Machine Learning,” International
Journal of Advanced Research in Computer and Communication Engineering, vol. 11, no. 5, May
2022.

[8] M. Nipas, A. G. Acoba, J. N. Mindoro, M. A. F. Malbog, J. A. B. Susa and J. S. Gulmatico, “Burned

Calories Prediction using Supervised Machine Learning: Regression Algorithm,” in 2022 Second
International Conference on Power, Control and Computing Technologies (ICPC2T), Raipur, India,
March 2022.

[9] R. S. Biyani and M. S. Nandini, “CALORIES PREDICTION BASED ON FOOD IMAGES,” International
Research Journal of Engineering and Technology (IRJET), vol. 7, no. 8, pp. 2122-2125, August 2020.

[10] K. Westerner, “Control of energy expenditure in humans,” European Journal of Clinical Nutrition,
vol. 71, pp. 340-344, 30 November 2016.

[11] “calories burnt _data,” Kaggle, 2022. [Online]. Available: https://www.kaggle.com/

datasets/Sandeep gauti/calories-burnt-data.

[12] K. K. Al-japery, T. Obafemi-Ajay, G. R, Olbricht and D. C. W. II, “Computational Learning

Approaches to Data Analytics in Biomedical Applications,” ACADEMIC PRESS, 2020, pp. 7-27.

[13] K.Nighania, “Various ways to evaluate a machine learning model’s performance,” Towards Data
Science, 30 December 2018. [Online]. Available: https://towardsdatascience. com/various-ways-to-
evaluate-a-machine-learning-models-performance-230449055f15.

[14] X. Wang, D. Fu, Y. Wang, Y. Guo and Y. Ding, “The XGBoost and the SVM-based prediction models
for bioretention cell decontamination effect,” Arabian Journal of Geosciences , vol. 669, no. 14, 6
April 2021.
 Abhishek Jaiswar
31031523011

[15] The Mifflin-St Jeor equation is a widely used tool to determine the resting metabolic rate [RMR].

The Mifflin-St Jeor Equation calculator is created by QxMD.