Computational Logic Building and Problem Solving in Seocho

1. Relationships Between Variables

Covariance and correlation describe how pairs of numeric variables move together:

• Positive — both rise or both fall together
• Negative — one rises while the other falls
• Near zero — little linear association

Both measures require numerical columns only—not categories or free text.

2. Covariance vs Correlation

• .cov() — direction in raw units; scale depends on feature units (−∞ to +∞), hard to compare strength across pairs
• .corr() — standardized to −1…+1; strength is easier to read; captures linear relationships only

3. Dataset Setup

import pandas as pd

df = pd.read_csv("ecommerce_orders.csv")
numerical_cols = ["Quantity", "Unit Price", "Discount", "Delivery Days", "Revenue"]
df_num = df[numerical_cols]
print(df_num.shape)

4. Covariance Matrix

cov_matrix = df_num.cov()
print(cov_matrix)

Diagonal entries are variances. Example patterns: Quantity–Revenue and Unit Price–Revenue positive; Discount–Revenue negative.

5. Correlation Matrix

corr_matrix = df_num.corr()
print(corr_matrix)
print(corr_matrix["Revenue"].sort_values(ascending=False))

Illustrative correlations with Revenue: Unit Price ≈ +0.67, Quantity ≈ +0.62, Discount ≈ −0.13, Delivery Days ≈ 0.005 (essentially uncorrelated).

6. Correlation ≠ Causation

Moving together is an association, not proof that one variable causes the other. Report relationships carefully without claiming causal effects unless experiments or domain evidence support them.