Documentation for the Macroframework Forecasting package.¶

This repository contains the Python code for the forecasting method described in:

Installation¶

To install the macroframe-foreacst package, run the following in the terminal/shell:

pip install macroframe-forecast

Quick start¶

The relevant import from macroframe-foreacst is MFF:

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from macroframe_forecast import MFF

# true data
df_true = pd.DataFrame({
    'var1': np.random.randn(30),  # 100 random values from normal distribution
    'var2': np.random.randn(30)
})
df_true['sum'] = df_true['var1'] + df_true['var2']

# input dataframe,
df = df_true.copy()
fh = 5
df.iloc[-fh:, 1:] = np.nan

# apply MFF
m = MFF(df, equality_constraints=['var1_? + var2_? - sum_?'])
df2 = m.fit()

# plots results
fig,axes = plt.subplots(3,1,sharey=True, figsize=(9,9))

axes[0].plot(df2['var2'], label='forecasted var2')
axes[0].plot(df_true['var2'], label='true var2')
axes[0].legend()

axes[1].plot(df2['sum'], label='forecasted sum')
axes[1].plot(df_true['sum'], label='true sum')
axes[1].legend()

axes[2].plot( df2['var1'] + df2['var2'] - df2['sum'], label='summation error')
axes[2].legend()

Index