Statsmodels

Statsmodels provides statistical modeling tools for time series and econometric analysis.

Time Series Decomposition

from statsmodels.tsa.seasonal import seasonal_decompose
import pandas as pd

# Ensure datetime index with frequency
df.index = pd.to_datetime(df.index)
df = df.asfreq('QE')  # Quarterly end frequency

# Perform seasonal decomposition
result = seasonal_decompose(
    df['current_account'],
    model='additive',  # or 'multiplicative'
    period=4  # quarterly data has period of 4
)

# Access components
trend = result.trend
seasonal = result.seasonal
residual = result.resid

# Plot decomposition
result.plot()
plt.savefig('/app/output/decomposition.png')

Extracting Decomposition Results

from statsmodels.tsa.seasonal import seasonal_decompose

result = seasonal_decompose(df['current_account'], model='additive', period=4)

# Create DataFrame with components
decomposition_df = pd.DataFrame({
    'observed': result.observed,
    'trend': result.trend,
    'seasonal': result.seasonal,
    'residual': result.resid
})

# Identify peak seasonal quarter
seasonal_by_quarter = df.groupby(df.index.quarter)['seasonal'].mean()
peak_quarter = f"Q{seasonal_by_quarter.idxmax()}"

Trend Analysis

from statsmodels.tsa.filters.hp_filter import hpfilter

# Hodrick-Prescott filter for trend extraction
cycle, trend = hpfilter(df['current_account'], lamb=1600)

# Determine trend direction
trend_slope = (trend.iloc[-1] - trend.iloc[0]) / len(trend)
trend_direction = "improving" if trend_slope > 0 else "deteriorating"

OLS Regression

import statsmodels.api as sm

# Time trend regression
X = sm.add_constant(range(len(df)))
y = df['current_account']

model = sm.OLS(y, X)
results = model.fit()

print(results.summary())
print(f"R-squared: {results.rsquared:.4f}")
print(f"Trend coefficient: {results.params[1]:.4f}")

Autocorrelation Analysis

from statsmodels.graphics.tsaplots import plot_acf, plot_pacf

fig, axes = plt.subplots(1, 2, figsize=(12, 4))

plot_acf(df['current_account'].dropna(), ax=axes[0], lags=12)
plot_pacf(df['current_account'].dropna(), ax=axes[1], lags=12)

plt.tight_layout()
plt.savefig('/app/output/acf_pacf.png')

Stationarity Testing

from statsmodels.tsa.stattools import adfuller

result = adfuller(df['current_account'].dropna())
print(f'ADF Statistic: {result[0]:.4f}')
print(f'p-value: {result[1]:.4f}')
print(f'Critical Values: {result[4]}')