Prophet › Example

import numpy as np
import pandas as pd

# Create sample monthly sales data (2020-2023)
np.random.seed(42)
dates = pd.date_range(start="2020-01-01", end="2023-12-31", freq="ME")

# Create realistic sales data with trend and seasonality
base_sales = 1000
trend_growth = 50  # Monthly growth
seasonal_amplitude = 200
noise_std = 50

sales_data = []
for i, date in enumerate(dates):
    # Trend component: linear growth
    trend = base_sales + trend_growth * i

    # Seasonal component: higher in December, lower in summer
    month = date.month
    if month == 12:  # December peak
        seasonal = seasonal_amplitude
    elif month in [6, 7, 8]:  # Summer dip
        seasonal = -seasonal_amplitude * 0.5
    else:
        seasonal = seasonal_amplitude * 0.3 * np.sin(2 * np.pi * month / 12)

    # Add noise
    noise = np.random.normal(0, noise_std)

    # Total sales
    total_sales = trend + seasonal + noise
    sales_data.append(total_sales)

# Create DataFrame
df = pd.DataFrame({"ds": dates, "y": sales_data})

print("Sample of our sales data:")
print(df.head(10))
print(f"\nData shape: {df.shape}")
print(f"Date range: {df['ds'].min()} to {df['ds'].max()}")

# Initialize and fit Prophet model
model = Prophet(
    yearly_seasonality=True,
    weekly_seasonality=False,  # We have monthly data
    daily_seasonality=False,
    changepoint_prior_scale=0.05,  # Controls trend flexibility
    seasonality_prior_scale=10.0,  # Controls seasonality strength
)

# Fit the model
model.fit(df)

# Make predictions
future = model.make_future_dataframe(
    periods=12, freq="ME"
)  # Forecast 12 months ahead
forecast = model.predict(future)

print("Forecast components for the last few months:")
print(forecast[["ds", "yhat", "trend", "yearly"]].tail(15))