Sales EDA Project — Data se Insight tak

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

BG = "#0d0d1c"
plt.rcParams.update({"figure.facecolor": BG, "axes.facecolor": BG,
    "axes.labelcolor": "#8a8aa3", "text.color": "white",
    "xtick.color": "#8a8aa3", "ytick.color": "#8a8aa3"})

df = pd.read_csv("sales.csv", parse_dates=["date"])

print(df.shape)            # rows x cols
print(df.head())
print(df.info())           # types + nulls
print(df.isna().sum())     # missing per column
print(df.duplicated().sum())   # duplicates
print(df.describe(include="all"))   # numeric + category summary

# Sanity: kya dates range me hain? amount >= 0?
print(df["date"].min(), df["date"].max())
print((df["amount"] < 0).sum())    # negative amounts = data error

# 1. Duplicates
df = df.drop_duplicates()

# 2. Missing
df["city"] = df["city"].fillna("Unknown")
df = df.dropna(subset=["amount"])     # amount hi nahi to row bekaar

# 3. Types
df["date"] = pd.to_datetime(df["date"])
df["amount"] = pd.to_numeric(df["amount"], errors="coerce")

# 4. Derived columns (analysis ke liye)
df["month"] = df["date"].dt.to_period("M").astype(str)
df["weekday"] = df["date"].dt.day_name()
df["unit_price"] = df["amount"] / df["qty"].replace(0, np.nan)

print("clean shape:", df.shape)
print(df.isna().sum().sum(), "total nulls left")

# Numeric: distribution
fig, axes = plt.subplots(1, 2, figsize=(12, 4.5))
df["amount"].hist(bins=40, ax=axes[0], color="#2dd4bf")
axes[0].set_title("Amount distribution", color="white")
df.groupby("city")["amount"].sum().sort_values().plot(
    kind="barh", ax=axes[1], color="#f0a500")
axes[1].set_title("City-wise total", color="white")
plt.tight_layout(); plt.show()

# Category: counts + share
print(df["category"].value_counts(normalize=True).round(3))

# Outliers: IQR method
q1, q3 = df["amount"].quantile([.25, .75])
iqr = q3 - q1
out = df[(df["amount"] < q1 - 1.5*iqr) | (df["amount"] > q3 + 1.5*iqr)]
print("outlier rows:", len(out))

# Month-wise trend (kya business badh raha hai?)
trend = df.groupby("month")["amount"].sum()
trend.plot(kind="line", marker="o", color="#f0a500", figsize=(10, 4))
plt.title("Monthly Revenue Trend", color="white")
plt.xticks(rotation=45); plt.tight_layout(); plt.show()

# Category x City heatmap
pt = df.pivot_table(index="city", columns="category",
                    values="amount", aggfunc="sum", fill_value=0)
sns.heatmap(pt, annot=True, fmt=".0f", cmap="viridis")
plt.title("City x Category Revenue", color="white")
plt.show()

# Top drivers
print(df.groupby("category")["amount"].agg(["sum", "mean", "count"])
        .sort_values("sum", ascending=False))

# Weekday effect
print(df.groupby("weekday")["amount"].mean().sort_values(ascending=False))

# Template - har insight: Finding + Evidence + Action
"""
INSIGHT 1 - Trend
Finding:  Revenue Jun se Oct tak lagatar upar (CAGR ~7%/month).
Evidence: month-wise sum, line chart.
Action:   Q4 stock aur staff pehle se ready rakho.

INSIGHT 2 - Concentration
Finding:  Top 2 cities = 62% revenue; bottom 3 = 8% only.
Evidence: city-wise sum (bar + share %).
Action:   Bottom cities me ya to marketing budget do ya
          operations consolidate karo.

INSIGHT 3 - Category margin proxy
Finding:  Electronics ka avg order value 3.4x Grocery ka,
          lekin order count sabse kam.
Evidence: category agg (sum/mean/count).
Action:   Electronics pe bundle offers (AOV already high),
          Grocery pe frequency campaigns (repeat chahiye).
"""