/fuel-consumption-prognosis

Primary LanguageJupyter Notebook

📈Fuel-Consumption-prediction - 📑Brief Summary

  • Used pandas to import and Sanitized the dataset.
  • performed EDA using various python libraries like matplotlib, Seaborn, Numpy
  • Build ML models for prediction using libraries Scikit-learn, etc.
  • Used Streamlit to deploy the M.L. model.

🪨Challenges and 🧠learnings-

Chi-squared analysis | Scikit-learn | Streamlit | Team Support

🤖Techstack-

Python | Pandas | Matplotlib | Seaborn| Scikit-learn | Streamlit | Tableau

Visit here for streamlit U.I. 👉 https://fuel-consumption-rating.streamlit.app/

👍Descriptive Explanation -

🥅Objective-

This project is for learning purposes. So, considering that got this data set from an Automobile dealer who deals in premium luxury cars and want to predict the Average fuel consumption of a vehicle basis on different parameters like- Vehicle class, Engine size, Transmission, Fuel type, etc. and to create a U.I. for User to find the consumption according to thier vehicle's Parameter.

📑E.D.A.-

  • Before E.D.A did the data wrangling.
  • After cleaning did Univariate & Bivariate Analysis to understand the features.

Univariate Analysis :

  • In this Visualization, we can observe how many vehicles are present in each Transmission type.

  • In this Visualization, we can observe that the maximum number of vehicles consume fuel in the range of 7 to 14 liters for 100km

for indepth Univariate analysis understanding 👉 https://github.com/manishhemnani06/FUEL_CONSUMPTION_ANALYSIS/blob/main/FUEL_CONSUMPTION_ANALYSIS_FILE.ipynb

Bivariate Analysis :

  • In this pair plot Visualization, we can observe different scatter plots giving the relation between all features.

  • In this bar graph Visualization, we can observe as the number of cylinders increases the fuel consumption is also increasing.

  • This Correlation Heatmap gives relation between all features of data set.

🤖M.L. Model-

- In this project, tried a total of three machine learning models linear regression, decision tree, and random forest. at last, we found that the linear model is best suitable as per the accuracy as well as the above analysis we found the data is full filling the linear regression assumptions.

🤖M.L. Model Deployment-

- Used pickle library to create a sas (Statistical Analysis System) file for model deployment.

💻User Interface-

- For user interface used streamlit library.

click on image to use U.I. :