/Bike-Rental-Prediction

Regression : Predicting the number of bikes to be rented based on environment and weather conditions

Primary LanguageJupyter Notebook

Bike-Rental_Prediction

Problem Statement

The objective of this Case is prediction of bike rental count on daily basis based on the environmental and seasonal settings.

Data

Our task is to build a regression model that will predict the bike rental on daily basis based on different environmental and seasonal settings. Given below is a sample dataset which we are using to predict the bike rental count.The varibale names are rather self explanatory.

day_data

Pre Processing

Distribution of "cnt" - The count of daily bikes rented

correaltion plot

Distribution of Number of Bikes-Rented by Month

monthly sales

Distribution of Number of Bikes-Rented by Season

season sales

Plot of "temp" -temperature Vs. "cnt" - The count of bikes rented

rplot-temp vs cnt

Plot of "dteday" - The dates Vs "cnt" - The count of bikes rented

rplot -dteday vs cnt

Outlier Analysis

We find outliers in windseed windspeed

Feature Selection

Based on the correlation plot, we remove atemp.

correlation plot

Model Selection

Multiple Linear Regression

Implemented multiple linear regression with an accuracy of 87%

Random Forest

Implemented Random Forest with an accuracy of 90%