Personal Finance

Tracking personal finances can be tedious. Either it takes a lot of time to keep everything well categorized as new transactions come in, or it is far from accurate because a tool tries to predict the categories for you. That might work fine for names such as “Wall Mart” or “Starbucks”, but your local bakery called “Morty’s Place” is not going to get picked up by the model. Many personal finance tools let you adjust these categories manually, but that is just as tedious as doing it from scratch.

With PersonalFinance I want to make it easier to manage your finances. By defining each category with appropriate keywords, you can be sure that the model will categorize transactions how you defined them. This is because it is not a generic model that is trained on a large dataset of transactions from all over the world. It is trained on your own data, which means that it will be able to categorize transactions that are specific to you. This results in Morty’s Place being correctly categorized as a Bakery.

So that you don’t need exact matches, the package uses the Levenshtein distance to determine how similar two strings are. This means that if you have a category called “Groceries” with the keyword “Supermarket” and a transaction comes in with the name “Rick’s Super Market”, it will still be categorized as “Groceries”. I purposely kept the Mumbo Jumbo to a minimum so that it stays clear why a transaction is categorized the way it is.

By doing most of these things through Python and Excel, you have complete freedom to decide what to do with the output. For example, you can use it to create your own personalized dashboards via any programming language or application such as Excel, PowerBI, Tableau, etc. I don’t want to bore you with custom dashboards that I tailored to myself just so that you can come to the conclusion that it isn’t a perfect fit for you.

Installation

Before installation, consider starring the project on GitHub which helps others find the project as well. Click the image to visit the repository and Star the project.

Installing PersonalFinance only requires the following:

pip install personalfinance -U

Then to use the features within Python use:

from personalfinance import Cashflow

cashflow = Cashflow()

This generates a configuration file, which you can supply again later with configuration_file='cashflow.yaml'. See below for more on each capability and what you can do with this file.

How-To Guides for PersonalFinance

The How-To guides below show how to use Personal Finance to gain insight into your own finances. They are written as Jupyter Notebooks, which you can open with the button below each description.

Getting Started

To get started, you need a configuration file that defines your transactions. It contains things such as the location of the datasets, the columns that hold e.g. the name, amount and date, and the categories and keywords used to categorize transactions. The configuration file is automatically downloaded on initialization.

To see an example, you can run the following code:

from personalfinance import Cashflow

cashflows = Cashflow(example=True)

cashflows.perform_analysis()

Before it does anything, it will download the example datasets as found here. This is only meant to show you how the functionality works. When you are ready to use it for your own cashflows, remove the example=True argument and supply your own configuration file. If you don’t have one yet, Cashflow() will supply one automatically. See the Notebooks here for an in-depth explanation.

The perform_analysis functionality does the following things:

1. It reads all the cashflow datasets based on the configuration file’s file_location parameter. This can be a single file, a selection of files or an entire folder. If chosen, it also applies a cost or income indicator when the numbers in your file are all positive (e.g. a column that says “Plus” or “Minus”).
2. It starts applying categorization based on the categories section in the configuration file. It uses Levenshtein distance to find matches that are closely related (e.g. ‘Tim’s Bakery’ and ‘Bakery’ would fit in the same category).
3. It generates multiple transactional and categorized overviews on a weekly, monthly, quarterly and yearly basis.
4. It generates an Excel file in which all of the results are displayed in a neat format based on the excel section of the configuration file. This is optional and can be disabled by setting write_to_excel to False.

See the resulting image for the file that is generated based on the example dataset:

You don’t have to continue in Excel if you are handy with Python, as all created datasets can also be accessed directly in Python through the related get functions, for example:

cashflows.get_period_overview(period='yearly')

Which returns:

And the following:

cashflows.get_transactions_overview(period='weekly')

Which returns:

These datasets make it possible to plot the spending pattern over time for each category. This can be done simply by selecting the column and using .plot() from Pandas, but it is also possible to create a larger overview as shown below:

import matplotlib.pyplot as plt

# Obtain the Quarterly Cashflow Overview
quarterly_cashflows = cashflows.get_period_overview(period='quarterly')

# Define the colormap
cmap = plt.get_cmap('tab20c')

# Create the figure and axes
fig, axes = plt.subplots(
    nrows=2,
    height_ratios=[6, 1],
    figsize=(30, 10))

# Plot the data per category
quarterly_cashflows.plot.bar(
    stacked=True,
    colormap=cmap,
    title="Quarterly Cashflow Overview",
    ax=axes[0])

# Calculate the totals
totals = quarterly_cashflows.sum(axis=1)

# Plot the totals
totals.plot.bar(
    color=['g' if x >= 0 else 'r' for x in totals],
    ax=axes[1])

# Format the plot by rotating labels and adjusting space
plt.xticks(rotation=45)
fig.subplots_adjust(wspace=0, hspace=0)

This returns the following plot: