longtrends

A package to download long-term Google Trends.

Introduction

Google Trends, downloadable by API using pytrends, limits the time period which can be downloaded using a single request. Each request is scaled between 0 and 100, making comparison between different time periods difficult. This package automatically downloads overlapping trends and rescales them, providing trend data across a long-term period.

Installation

pip install longtrends

Requirements

Requires pytrends, installed automatically with pip.

Quick Start

from longtrends import LongTrend
from datetime import datetime

keyword = 'suncream'

# Create LongTrend object
longtrend = LongTrend(
                      keyword=keyword,
                      start_date=datetime(2018, 1, 1),
                      end_date=datetime(2022, 3, 31))        # use verbose=True for print output
# Build long-term trends
lt_built = longtrend.build()

# Plot
lt_built.plot(title=f"Google Trends: {longtrend.keyword}", figsize=(15, 3))

Under the hood

First, longtrends downloads overlapping trends.

from longtrends import rescale_overlaps, get_overlapping_trends, rescaled_longtrend
from datetime import datetime
import pandas as pd

keyword = 'suncream'

overlapping = get_overlapping_trends(
                                keyword=keyword,
                                start_date=datetime(2018, 1, 1),
                                end_date=datetime(2022, 3, 31),
                                verbose=True)

pd.concat(overlapping, axis=1).plot(figsize=(15,3), legend=False)

Next, i+1th overlap is rescaled to ith overlap.

rescaled = rescale_overlaps(overlapping)
pd.concat(rescaled, axis=1).plot(figsize=(15,3), legend=False)

Finally, a single long-term trend is picked.

rescaled = rescaled_longtrend(rescaled)
rescaled.plot(figsize=(15,3), title='Rescaled long-term trend')

Disclaimer

This is not an official or supported product. It is provided without warranty under MIT license.