Does one series lead another? in Python

The Granger test asks whether past x improves a forecast of y.

import numpy as np
from statsmodels.tsa.stattools import grangercausalitytests

rng = np.random.default_rng(16)
n = 300
leader = rng.normal(size=n)
follower = np.zeros(n)
for i in range(2, n):
    follower[i] = 0.5 * leader[i - 2] + rng.normal(scale=0.5)

data = np.column_stack([follower, leader])
results = grangercausalitytests(data, maxlag=3, verbose=False)

for lag, result in results.items():
    pvalue = result[0]["ssr_ftest"][1]
    print(f"lag {lag}: p={pvalue:.5f}")

How it works

1. The data goes in as two columns, y first.
2. maxlag bounds how far back to look.
3. It tests prediction, which is not the same as causation.

Keywords and builtins used here

• as
• for
• print
• range

The run, in numbers

Lines

16

Characters to type

465

Tokens

142

Three-star pace

115 tpm

At the three-star pace of 115 tokens a minute, this run takes about 74 seconds.

Type this snippet

Step 6 of 6 in Time series, step 18 of 19 in Statistics with statsmodels.

← Previous Next →