[bug] sparsity returning incorrect values

[ngupta23]

What happened + What you expected to happen

Sparsity is not being calculated correctly in the code. See below. The data clearly has 0 values and the expected sparsity is 0.5, but the code is returning a value of 0.

Versions / Dependencies

Name: tsfeatures Version: 0.4.5

Reproduction script

import pandas as pd
data = {'y': [0, 23, 0, 45, 0, 19, 0, 7, 34, 0]}
df = pd.DataFrame(data)
df["unique_id"] = "A"
from datetime import datetime, timedelta
start_date = datetime(2024, 1, 1)
df['ds'] = [start_date + timedelta(weeks=i) for i in range(len(df))]
df

# Extract features
from tsfeatures import sparsity
tsfeatures(df, freq=52, features=[sparsity])

# Expected result
np.mean(df["y"] == 0)  # 0.5

Issue Severity

Medium: It is a significant difficulty but I can work around it.

[ngupta23]

This is because the scale argument is set to True by default. To calculate the correct value, set scale to False

tsfeatures(df, freq=52, features=[sparsity], scale=False)

[ngupta23]

Looks like an enhancement request is already in place: #40