Data Visualization on Old Cars Price Dataset

About Dataset:

• An Indian dataset on Old Car Prices.
• Data extracted from https://www.machinehack.com/course/predicting-the-costs-of-used-cars-hackathon-by-imarticus/
• Original dataset: 'Data_Train.xlsx'
• The dataset originally used on the wesbite has been cleaned by me, which is the one included in the project as 'car_clean.csv'
• Cleaned Dataset has 6019 data points and 14 features.
• The data_viz_slides.html is the final presentation consisting of tutorial.

Packages used:

• numpy
• pandas
• matplotlib
• seaborn

Some common plots used:

• Histograms
• Bar graphs
• Boxplots
• Violin Plots
• Barplots
• Clustered Bar plots
• Scatterplots
• Heatmaps

Conclusions:

Univariate:

• Top cities of old cars: Mumbai, Hyderabad, Kochi, Coimbatore. least Ahmedabad
• Kilometers Driven is almost uniform distribution
• Diesel and Petrol constitutes more than 80% of cars
• Most cars are manual
• Price of Car: first > Second > third > Fourth & Above
• log of prices is normally ditributed. ie price is highly right skew
• Top 4 cars: Maruti, Hyunday, Honda, toyota constitute more than 80% of cars

Bivariate:

• Order of Mileage: CNG > LPG, Diesel > Petrol
• Mileage of ELectric Vehicles are NaN
• LPG has high variability
• CNG is highly left skew
• Highest average price of car in Coimbatore and lowest in Kolkata
• Price has a direct correlation with Price
• Electric and Diesel cars are expensive, while CNG and LPG are cheaper
• Price of Automatic Cars is SIGNIFICANTLY high than Manual cars
• Engine_CC and Price: slightly +ve correlation
• Power_bhp and price: good +ve correlation
• Power_bhp and Engine_CC: Highly =ve correlation
• Kilometers Driven and Year: Good -ve correlation

Multivariate

• LPG and CNG lie have lower Engine_CC and Power_bhp
• Only Petrol cars lie in the higher Power_bhp and Engine_CC range
• In Manual vs Automatic cars, the trend of Petrol, Electric and Disel car prices is completely reversed