Empowering the Automotive Industry with Data Analytics.

Use Data to Get to know about your customers and their needs.

The real benefit of data analytics is its ability to inform business decisions that fast track growth. After analyzing the data, you should be able to identify trends. These trends reveal actionable information that can be used to make changes to your website, app, product, etc. to optimize performance.

And we are happy to provide the accurate Data Analysis that will help you rule the Automotive Sector by understanding customers.

A powerful tool such as Data Analytics has the potential to take the automotive industry to new heights by helping them in making calculated decisions with the help of analytical proofs and research.

This is the dataset that I considered for performing the Data Analysis.

Dataset:

	Unnamed: 0	Make	Model	Variant	Ex-Showroom_Price	Displacement	Cylinders	Valves_Per_Cylinder	Drivetrain	Cylinder_Configuration	Emission_Norm	Engine_Location	Fuel_System	Fuel_Tank_Capacity	Fuel_Type	Height	Length	Width	Body_Type	Doors	...	Rear_Center_Armrest	iPod_Compatibility	ESP_(Electronic_Stability_Program)	Cooled_Glove_Box	Recommended_Tyre_Pressure	Heated_Seats	Turbocharger	ISOFIX_(Child-Seat_Mount)	Rain_Sensing_Wipers	Paddle_Shifters	Leather_Wrapped_Steering	Automatic_Headlamps	Engine_Type	ASR_/_Traction_Control	Cruise_Control	USB_Ports	Heads-Up_Display	Welcome_Lights	Battery	Electric_Range
535	535	Mahindra	Bolero Power Plus	Lx	Rs. 7,49,192	1493 cc	4.0	2.0	RWD (Rear Wheel Drive)	In-line	BS IV	Front, Transverse	Injection	60 litres	Diesel	1880 mm	3995 mm	1745 mm	SUV	5.0	...	Yes	NaN	NaN	NaN	NaN	NaN	Yes	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
62	62	Maruti Suzuki	Celerio X	Vxi (O)	Rs. 4,81,074	998 cc	3.0	4.0	FWD (Front Wheel Drive)	In-line	BS IV	Front, Transverse	Injection	35 litres	Petrol	1560 mm	3600 mm	1600 mm	Hatchback	5.0	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
709	709	Toyota	Innova Crysta	Touring Sport 2.7 Vx 7 Str	Rs. 18,92,000	2393 cc	4.0	4.0	RWD (Rear Wheel Drive)	In-line	BS VI	Front, Longitudinal	Injection	55 litres	Petrol	1795 mm	4735 mm	1830 mm	MUV	5.0	...	Cup Holders	Yes	NaN	Yes	NaN	NaN	Yes	Yes	Yes	NaN	Yes	Yes	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
732	732	Jeep	Compass	2.0 Limited Plus 4X4 At	Rs. 24,99,000	1956 cc	6.0	4.0	AWD (All Wheel Drive)	V	BS 6	Front, Longitudinal	Injection	60 litres	Diesel	1640 mm	4395 mm	1818 mm	SUV	5.0	...	Cup Holders	Yes	Yes	NaN	NaN	NaN	Yes	Yes	Yes	NaN	Yes	Yes	NaN	Yes	NaN	NaN	NaN	NaN	NaN	NaN
1160	1160	Porsche	Macan	S	Rs. 85,03,000	2995 cc	6.0	4.0	AWD (All Wheel Drive)	In-line	BS IV	Front, Longitudinal	Injection	65 litres	Petrol	1624 mm	4696 mm	1923 mm	SUV	5.0	...	Cup Holders	Yes	Yes	Yes	NaN	All	Yes	Yes	Yes	Yes	Yes	Yes	NaN	Yes	Yes	NaN	NaN	NaN	NaN	NaN
11	11	Datsun	Redi-Go	1.0 S Amt	Rs. 4,37,065	999 cc	3.0	4.0	FWD (Front Wheel Drive)	In-line	BS IV	Front, Transverse	Injection	28 litres	Petrol	1541 mm	3429 mm	1560 mm	Hatchback	5.0	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
585	585	Mahindra	Xuv300	1.2 W4	Rs. 8,30,127	1197 cc	NaN	NaN	RWD (Rear Wheel Drive)	In-line	BS 6	Front, Transverse	Injection	42 litres	Petrol	1617 mm	3995 mm	1821 mm	SUV	5.0	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
978	978	Kia	Seltos	Htx Plus 1.5 Diesel	Rs. 15,34,000	1493 cc	NaN	NaN	FWD (Front Wheel Drive)	In-line	BS 6	Front, Longitudinal	Injection	60 litres	Diesel	1645 mm	4315 mm	1800 mm	SUV	5.0	...	Cup Holders	Yes	NaN	Yes	NaN	Driver	Yes	Yes	Yes	Yes	Yes	Yes	NaN	NaN	Yes	NaN	NaN	NaN	NaN	NaN

8 rows × 141 columns

The Dataset Statistics were as follows:

After this we can choose the desired features to perform Detailed Analysis.

First we Check the price distribution, we will use both normal and log scales due to the huge difference in prices.
Plot shows the price distribution of various cars.

Here I have plotted a box plot  so that we can clearly examine the variance that we observed in the prices.

Seems that there is a lot of outliers that form a very different type(s) of cars or to be more precise there are very different categories in the market.

Plot here shows the most commmon cars sorted as per thir body types.

Conclusion: SUV's Sedans and hatchbacks seems to be the dominating car types.

Conclusion: Indian market is favourable for SUVs, sedans, and Hatchback.

The box plot below shows that how price of the car varies with respect to its body type.

Conclusion: It's Clear from the plot above that Car body type strongly affect the price.

The plot below shiows the engine fuel type of the cars.

Here is a pie chart depiction of it.

Conclusion: Most cars seems to be run on gas or Diesel rather than other fuels which is not a good sign for the environment.

This data is going to change because electric vehicles have arrived in India.

Now let's see that what companies holds control over Indian market ( I am saying Indian because of the choice of our dataset and this can be applied to any dataset)

The plot below show the Top Car Making Companies in India.
As we have considered the Indian Cars Dataset here we are depicting that in a graph,
same can be done with other datasets as well

Distribution of cars by engine size.

Conclusion: Seems like most of cars have engine size in the 1000:2000cc range.(More frequently purchased car types have engine size in this range)

Next We checked the Horsepower of the cars.

Plot showing the relation horsepower and price considering different body types.

Conclusions: Horsepower of car seems to be highly related to car price.

But car body type seems a little bit blurry.

But hatchbacks seems to be the body type with the least horsepower and price.

Next we are looked into the relation between Mileage and price.

<Figure size 720x576 with 0 Axes>

Conclusion: looks like expensive cars tend to have worse mileage.

checking the overall correlation of between variables and each other.

First we make a pearson correlation grid.

Now, checking an extensive scatter plot grid of more numerical variable to investigate the realtion in more detail.

Seems there are a lot of multicollinearity between variables.

I plotted a 3D scatter plot to check for obvious clusters with main features as price horsepower and mileage.

Clustering the market needs a lot of effort as the separation of clusters is not that obvious.

But It's now clear that we have to look to many dimensions in order to cluster the market,as the more features we explore the harder it's to cluster the market.

These dimensions affect the decision of the buyers not to mention it also precvied totally different due to the very different mental models of buyers,in other words, price horsepower and mileage are not everything,some buyers would like to have a long wheel base car, some would like to have wider car all of the previous features, and more, strongly affect the buyer' decisions.

This means that two car can have very similar price and milage but one is a van with lots of space and the other is just a four doors sedan, these two cars are precieved as two different categories in the automotive industry so space "length width and height of the car" can also be a vital factor.

So, a three dimensional representation won't tell everythings,

so thats why we will try to consider clustering to use the very different features associated with each car.

Clustering:

Cluster analysis or clustering is the task of grouping a set of objects in such a way that objects in the same group (called a cluster) are more similar (in some sense) to each other than to those in other groups (clusters). It is a main task of exploratory data analysis, and a common technique for statistical data analysis, used in many fields, including pattern recognition, image analysis, information retrieval, bioinformatics, data compression, computer graphics and machine learning.

Cluster analysis itself is not one specific algorithm, but the general task to be solved. It can be achieved by various algorithms that differ significantly in their understanding of what constitutes a cluster and how to efficiently find them. Popular notions of clusters include groups with small distances between cluster members, dense areas of the data space, intervals or particular statistical distributions. Clustering can therefore be formulated as a multi-objective optimization problem.

The type of clustering used here is k-means clustering.

Introduction to K-Means Algorithm:

The K-means clustering algorithm computes centroids and repeats until the optimal centroid is found. It is presumptively known how many clusters there are. It is also known as the flat clustering algorithm. The number of clusters found from data by the method is denoted by the letter ‘K’ in K-means.

In this method, data points are assigned to clusters in such a way that the sum of the squared distances between the data points and the centroid is as small as possible. It is essential to note that reduced diversity within clusters leads to more identical data points within the same cluster.

Here I am considering the example of corolla and clustering them in order to give a better competitor analysis. Similarly we can consider any other cluster group and analyze the market and models competitors in arather better way.

df = df[df.price < 60000]

num_cols = [ i for i in df.columns if df[i].dtype != 'object']

km = KMeans(n_clusters=8, n_init=20, max_iter=400, random_state=0)
clusters = km.fit_predict(df[num_cols])
df['cluster'] = clusters
df.cluster = (df.cluster + 1).astype('object')
df.sample(5)

	make	model	car	variant	body_type	fuel_type	fuel_system	type	drivetrain	displacement	cylinders	mileage	power	torque	fuel_tank	height	length	width	doors	seats	wheelbase	airbags	price	cluster
1126	Hyundai	Creta	Hyundai Creta	1.6 Crdi Sx (O)	SUV	Diesel	Injection	Manual	FWD (Front Wheel Drive)	1582	4	19.67	126.25	260	55.0	1630.0	4270.0	1780.0	5	5	2590.0	6	21609	8
1115	Skoda	Rapid	Skoda Rapid	Onyx Mt Diesel	Sedan	Diesel	Injection	Manual	FWD (Front Wheel Drive)	1498	4	21.13	108.50	250	55.0	1466.0	4413.0	1699.0	4	5	2552.0	2	16220	4
573	Hyundai	Verna	Hyundai Verna	1.6 Crdi Sx	Sedan	Diesel	Injection	Manual	FWD (Front Wheel Drive)	1582	4	23.90	126.25	260	45.0	1445.0	4440.0	1729.0	4	5	2600.0	2	16415	4
190	Ford	Aspire	Ford Aspire	1.5 Tdci Titanium Plus	Sedan	Diesel	Injection	Manual	FWD (Front Wheel Drive)	1498	4	26.10	98.63	215	40.0	1525.0	3995.0	1704.0	4	5	2490.0	6	12073	5
220	Toyota	Glanza	Toyota Glanza	V	Hatchback	Petrol	Injection	Manual	FWD (Front Wheel Drive)	1197	4	21.01	80.88	113	37.0	1540.0	3995.0	1745.0	5	5	2520.0	2	10614	1

Now we check some scatter plots but with adding clusters

price vs power

Conclusion: We can see the the clusters are strongly affected by the price with clear speration between clusters but it's kind of blurry when it comes to power.

power vs mileage

Conclusion: But yet we can see that clusters speration in power is stronger than mileage which almost have no separation of clusters>

Engine size vs Fuel tank

I made an interactive 3D scatter plot of price power, and mileage using also clusters.

3D plot:

checking the average prices of each cluster.

checking average prices of each cluster

Conclusion: As shown in the scatter plits earlier there is a clear seperation of clusters when it comes to prices.

checking that how many cars exist in each cluster.

Number of cars in each cluster.

Conclusion: We can generally say that even if clusters generated are not determinant yet we can see that they still can be useful.

Finding the potential strategic group

first we find the cluster of the Toyota Corolla (and its variants)

We found that the cluster of the corolla is cluster 1 and also cluster 5 we can now search these clusters and check what is intersing about it.

First we check a sample of these clusters

Here is a more interactive chart that shows cars prices including all variants (with maximum and minimum value of each car).

A more interactive chart that shows cars prices including all variants (with max and min value of each car).

Count of each body type in the targeted cluster
(here we have taken the corolla cluster for performing our analysis).

seems like there are too many SUV's in the Toyota Clusters, should that be important?

With clustering there are too many variable taken in considration which are hard to be traced by normal methods. The clusters generated by the KMeans model can be used to identify what is the strategic group that form a strong competition to the company products in the market it also show the close clusters to this group which also can be put in considration in some cases.

Problem with clustering:

As exciting and tempting as it's to use clustering to produce strategic groups it worth mentioning that the clustering process itself is a little bit ambigous and features contribution to the clustering process can't be easily explained so the overall interpretability of the model forms a challenge.

Conclusion:

clustering may be not determinant but it can be used to augment the management decision by using it side by side with human intuition to come out with the right strategic group.

Conclusion from our Data Analyis:

1.There are a lot of outliers that form very different types (s) of cars or to be more precise there are very different categories in the market.

2.SUVs, Sedans, and hatchbacks seem to be the dominating car types.

3.The Indian market is favorable for SUVs, sedans, and Hatchback.

4.It's clear from the plot above that Car body type strongly affects the price.

5.Most cars seem to be run on gas or Diesel rather than other fuels which is not a good sign for the environment.This data is going to change because electric vehicles have arrived in India.

6.Seems like most cars have the engine sizes in the 1000:2000cc range.

7.The horsepower of the car seems to be highly related to car price. But hatchbacks seem to be the body type with the least horsepower and price.

8.Looks like expensive cars tend to have worse mileage.

9.But It's now clear that we have to look to many dimensions in order to cluster the market, as the more features we explore the harder it's to cluster the market.

10.These dimensions affect the decision of the buyers not to mention it is also perceived totally different due to the very different mental models of buyers, in other words, price horsepower and mileage are not everything, some buyers would like to have a long wheelbase car, some would like to have a wider car all of the previous features, and more, strongly affect the buyer' decisions.This means that two-car can have very similar prices and milage but one is a van with lots of space and the other is just a four doors sedan, these two cars are perceived as two different categories in the automotive industry so space "length width, and height of the car" can also be a vital factor.