65. How to Forecast Sales with Time Series in Python (pandas and sklearn)

65.2 What Is a Time Series (Trend, Season, Noise)

65.2 Time series म्हणजे काय (trend, season, noise)

By Ravindra Bagale, trainer · LinkedIn · Last updated: 31 October 2026

फक्त एक formula शिकून Diwali jump समजत नाही. Trend, season, noise shape सांगतात.

हे का शिकायचं?

Why. If you only learn one formula, you miss why Diwali months jump. Trend, season and noise explain the shape.

हे काय आहे?

What.

• Trend — slow rise or fall over years.
• Season — a pattern that repeats (festivals, weekends, Q4).
• Noise — leftover bumps that have no clean story.

कसं करायचं?

How.

Worked example A — Rising trend in IT SaaS (₹).

Dataset note: fictional Nimbus Soft (Pune) quarterly MRR. 6 rows. Columns: quarter, mrr_inr.

import pandas as pd

saas = pd.DataFrame({
    "quarter": ["2025-Q1", "2025-Q2", "2025-Q3", "2025-Q4", "2026-Q1", "2026-Q2"],
    "mrr_inr": [850000, 920000, 990000, 1100000, 1180000, 1250000],
})
print(saas)
print("Change Q1→Q2 2026:", saas["mrr_inr"].iloc[-1] - saas["mrr_inr"].iloc[-2])

Output you should see.

  quarter  mrr_inr
0 2025-Q1   850000
1 2025-Q2   920000
2 2025-Q3   990000
3 2025-Q4  1100000
4 2026-Q1  1180000
5 2026-Q2  1250000
Change Q1→Q2 2026: 70000

याचा अर्थ. Clear upward trend. A flat “last value” forecast still works short-term, but lag + trend features help more.

Worked example B — Festival season in Mauli sales.

import pandas as pd

df = pd.read_csv("mauli_monthly_sales.csv", parse_dates=["month"])
df["month_num"] = df["month"].dt.month
print(df.groupby("month_num")["sales_inr"].mean().round(0).astype(int))

याचा अर्थ. Some calendar months average higher (festival stretch). That is season. Seasonal naive uses last year’s same month.

Worked example C — Noise in hospital beds (Bengaluru).

Dataset note: fictional Surya Care daily beds occupied. 7 rows. Columns: day, beds_occupied.

import pandas as pd

hosp = pd.DataFrame({
    "day": pd.date_range("2026-10-20", periods=7, freq="D"),
    "beds_occupied": [142, 148, 151, 149, 155, 160, 158],
})
print(hosp)
print("Day-to-day change:", hosp["beds_occupied"].diff().tolist())

Output you should see.

         day  beds_occupied
0 2026-10-20            142
1 2026-10-21            148
2 2026-10-22            151
3 2026-10-23            149
4 2026-10-24            155
5 2026-10-25            160
6 2026-10-26            158
Day-to-day change: [nan, 6.0, 3.0, -2.0, 6.0, 5.0, -2.0]

याचा अर्थ. Small up/down moves are often noise. Do not overfit them.