Skip to content

Latest commit

 

History

History
65 lines (48 loc) · 2.34 KB

README.md

File metadata and controls

65 lines (48 loc) · 2.34 KB
Raw

Fuel Efficiency prediction

with Root mean sqauared error of 2.5 miles per gallon

Deployment link - Make Prediction using various car features

Integrating web application with database to collect data for prediction using cloud database service 'Deta'

Source:

This dataset was taken from the StatLib library which is maintained at Carnegie Mellon University. The dataset was used in the 1983 American Statistical Association Exposition.

link - Auto mpg (UCI machine learning repository)

Data Set Information:

This dataset is a slightly modified version of the dataset provided in the StatLib library. In line with the use by Ross Quinlan (1993) in predicting the attribute "mpg", 8 of the original instances were removed because they had unknown values for the "mpg" attribute. The original dataset is available in the file "auto-mpg.data-original".

Attribute Information:

  1. mpg : fuel efficiency in miles per gallon (continuous)
  2. cylinders : Number of cylinders in a car (multi-valued discrete)
  3. displacement : in cubic inches (continuous)
  4. horsepower : (continuous)
  5. weight : weight of car in pounds (continuous)
  6. acceleration : (continuous)
  7. model year : (multi-valued discrete)
  8. origin : 1,2,3 (multi-valued discrete)
  9. car name : string (unique for each instance)

Total 3 multivalued discrete and 5 continuous attributes

Objective:

Prediction of city-cycle fuel consumption in miles per gallon given technical aspects & vehicle information

Alt Text

Assumptions:

origin is taken as 1: USA 2:Europe 3: Asia for convenience

Final Predictive Model

Multiple linear Regression

Kaggle - https://www.kaggle.com/code/hharvind/fuel-efficiency-prediction-in-mpg

Skills Aquired-

  1. Exploratory Data Analysis
  2. Data pre-processing
  3. KNN-Imputer
  4. Linear Regression
  5. L1 & L2 Regularization
  6. Polynomial Linear Regression
  7. KNearestNeighbour Regression
  8. Automatic Report generation (Functional Programming)

Hurdles

  1. Automobile Domain
  2. Treating outliers when Less amount of Data available
  3. High Multi-collinearity
  4. Deployment and integration with Database

Liabraries used

  1. Numpy
  2. Pandas
  3. Scikit-learn
  4. Streamlit (Deployment)
  5. Deta