EV Analysis: Exploratory Data Analysis with 'Price' as the Target Variable

1.Categorical Variable Analysis

General Observations - Brand:

• Porsche has the highest mean and median prices, suggesting it's a more premium brand with a wide range of prices. The high standard deviation indicates significant price variation across different models.
• Mercedes also has a wide range of prices, but with a lower median than Porsche, suggesting it offers more mid-range options alongside its luxury models.
• Opel and Peugeot are more affordable brands, with lower means and standard deviations, indicating more consistent, tightly clustered prices.
• Audi sits somewhere in the middle, with moderate variation in prices but with some high-end vehicles.

General Observations - Drive:

• All Wheel Drive vehicles have the highest mean price (€109,591) and the widest price range, with a maximum of €191,096. The high standard deviation indicates substantial variation in prices across models with this drive type. The median (€111,679) is close to the mean, indicating a relatively symmetric distribution.
• Rear Wheel Drive vehicles have a mean price of €83,610, with a median higher than the mean (€89,351), suggesting a slight left skew, meaning that the more expensive models may be pulling the median higher. The standard deviation (€20,285) shows moderate price variation.
• Front Wheel Drive vehicles are the most affordable, with the lowest mean (€54,008) and median (€53,640). They also have the smallest standard deviation (€10,480), indicating that prices are tightly clustered and consistent across models.
• In general, All Wheel Drive vehicles tend to be the most expensive and varied in price, while Front Wheel Drive vehicles are more affordable with less price variation.

ANOVA Results

1. ANOVA Result for Brands:

• F-statistic: 23.84
• p-value: 1.38e-13 (or 0.0000000000001376)

2. ANOVA Result for Drive Types:

• F-statistic: 43.77
• p-value: 3.36e-14 (or 0.0000000000000336)

Interpretation:

1. Brands:

• The F-statistic of 23.84 indicates a significant difference between the means of prices across different brands.
• The p-value of 1.38e-13 is extremely low, much lower than the common significance level of 0.05. This suggests that there is strong evidence to reject the null hypothesis (which states that all brands have the same average price).
• Conclusion: The price of electric vehicles varies significantly between different brands.

2. Drive Types:

• The F-statistic of 43.77 suggests that the difference in the means of prices between different drive types is even more pronounced.
• The p-value of 3.36e-14 is also much lower than 0.05, indicating that the null hypothesis (that prices are the same for all drive types) can be rejected.
• Conclusion: There is a significant difference in prices between different drive types (e.g., All Wheel Drive vs. Front Wheel Drive).

2. Numerical Variable Analysis

3. Preliminary Multiple Linear regression Predict Price

• Question: How sensitive is the price to all variables?
• Objective: Determine the factors influence price.

Key Takeaways from the OLS Regression Results

1. R-squared = 0.821:

   • The model explains about 82.1% of the variance in the dependent variable (PriceinGermany).
   • This indicates a strong model fit, though not perfect.

2. Adjusted R-squared = 0.815:

   • The adjusted R-squared accounts for the number of predictors in the model.
   • It's still a good value, indicating the model generalizes well beyond the training data.

3. F-statistic and p-value (Prob > F):

   • F-statistic = 137.1: The independent variables collectively have a significant relationship with the dependent variable.
   • p-value (4.05e-95): The very small p-value indicates the overall model is highly significant.

4. Significant Coefficients (p < 0.05):

   • x1 (Usable Battery (kWh)): Positive coefficient (964.67), suggesting that an increase in battery capacity is associated with an increase in price.
   • x2 (Acceleration (sec)): Positive coefficient (2020.36), meaning faster acceleration (lower time in seconds) leads to higher prices.
   • x3 (TopSpeed (km/h)): Positive coefficient (351.16), meaning higher top speeds lead to higher prices.
   • x8 (Drive_encoded): Positive coefficient (5904.15), showing that a higher encoding (likely all-wheel drive) increases price significantly.
   • x9 (Brand_encoded): Positive coefficient (0.6994), meaning that brands with a higher average price (target encoded) are associated with higher vehicle prices.

5. Non-Significant Variables (p > 0.05):

   • x4 (Range (km)), x5 (Efficiency (Wh/km)), x6 (FastChargeSpeed (km/h)), x7 (NumberofSeats):
     • These variables do not have a statistically significant impact on price.

6. Multicollinearity:

   • Condition Number = 1.84e+06: This large condition number indicates potential multicollinearity.
   • Multicollinearity can cause unreliable coefficient estimates and inflate standard errors.

7. Durbin-Watson = 1.300:

   • This value tests for autocorrelation in the residuals. Since 1.300 is below 2, there may be some positive autocorrelation in the residuals, though it is not severe.

8. Skew and Kurtosis:

   • Skew = 0.593: Indicates some asymmetry in the distribution of residuals.
   • Kurtosis = 4.823: Indicates heavier tails in the residual distribution (more outliers).

---

Next Steps:

1. Address Multicollinearity:

   • Compute the Variance Inflation Factor (VIF) and remove or combine variables with high multicollinearity (VIF > 10).

2. Recheck Model Fit:

   • After addressing multicollinearity, refit the model to see if the coefficients and their significance improve.

3. Consider Transformations:

   • Consider log transformations to handle skewness and make the residuals more normally distributed.

4. Residual Analysis:

   • Perform residual diagnostics to check if the residuals are normally distributed and if there are any patterns in the residuals.