Cashflow tracker with AI
In this post, we aim to guide you through a step-by-step approach to gaining insights from your financial situation with the assistance of AI, particularly ChatGPT. In this brief post, we will demonstrate how to utilize the free GPT-4 to categorize your monthly statement into income, savings, and expenses. We will provide the source code for free, allowing you to customize it according to your specific use case.
Step 0: Get your monthly bank statement
In this step, you simply need to access your online banking and request a monthly or period statement. The best option is a CSV file, as it facilitates more accurate parsing. However, if your bank does not provide that option, you can opt for a PDF format. Then, try copying the contents, including dates, prices, and descriptions, to a text file. To maximize the effectiveness of this tracker, it is recommended to use your checking account, which encompasses both income and expenses related to your credit card or other payments. However, you can choose either account.
Step 1: access free GPT-4
Wait! Is it possible? Yes, you can access the free version of GPT-4 on the Bing website of Microsoft Edge. To do so, simply download Microsoft Edge, then type bing.com. There is a chat option at the top that you can select to engage with GPT-4. For our case, the best option is to choose the more precise option.
Step 2: Prompt engineering to Jsonify our statement
For this step, you can use this specific prompt and add the statement to the end of it delimited by ###.
Step 3: Use python to plot
Finally in the last step, you save the json file generate by the GPT and put inside your python working directory. More detail in our Github page.
Click to show the whole code
cashflow = []
incomes_plot = {}
expenses_plot = {}
savings_plot = {}
for month in data:
# Income
incomes = data[month]["Income"]
income_values = list(incomes.values())
# compute the sum for the dictionary values in the list
income_values = [sum(i.values()) for i in income_values]
income_category = list(incomes.keys())
total_incomes = sum(income_values)
print(income_category)
# put all the income values in income_plot dictioanry for each category, on each month.
# if specific category is not in the dictionary, add it to the dictionary. put zero for previous months
# if some categories does not have values, put zero for that month
for i in income_category:
if i not in incomes_plot:
incomes_plot[i] = [0] * len(data)
incomes_plot[i][list(data).index(month)] = income_values[income_category.index(i)]
# Expenses
expenses = data[month]["Expenses"]
expenses_values = list(expenses.values())
expenses_values = [sum(i.values()) for i in expenses_values]
expense_category = list(expenses.keys())
total_expenses = sum(expenses_values)
# put all the expenses values in expenses_plot dictioanry for each category, on each month.
for i in expense_category:
if i not in expenses_plot:
expenses_plot[i] = [0] * len(data)
expenses_plot[i][list(data).index(month)] = expenses_values[expense_category.index(i)]
# Savings
savings = data[month]["Savings"]
savings_values = list(savings.values())
savings_values = [sum(i.values()) for i in savings_values]
savings_category = list(savings.keys())
total_savings = sum(savings_values)
# put all the savings values in savings_plot dictioanry for each category, on each month.
for i in savings_category:
if i not in savings_plot:
savings_plot[i] = [0] * len(data)
savings_plot[i][list(data).index(month)] = savings_values[savings_category.index(i)]
# Calculate the cashflow for each month
cashflow.append(total_incomes + total_expenses + total_savings)
# Print the cashflow for each month
print(cashflow)
# plot the cashflow over time
plt.figure(figsize=(10, 5))
plt.plot(list(data), cashflow, label='Cashflow')
# plot the summation of the cashflow over time
plt.plot(list(data), [sum(cashflow[:i]) for i in range(1, len(cashflow)+1)], label='Cumulative Cashflow')
plt.xlabel('Month')
plt.legend(loc='best')
# let's move the Bill Payement, MB-QUESTRADE from expenses to savings
# remove the Bill Payement, MB-QUESTRADE from expenses_plot dictionary
invest = expenses_plot.pop('Bill Payment')
# add the Bill Payement, MB-QUESTRADE to savings_plot dictionary and make the values negative
# savings_plot['Bill Payment'] = invest
# make all the values in the saving_plot dictionary negative
# savings_plot = {k: [-i for i in v] for k, v in savings_plot.items()}
# plot the incomes_plot, expenses_plot, savings_plot
plt.figure(figsize=(10, 5))
plt.stackplot(list(data), incomes_plot.values(), labels=incomes_plot.keys())
plt.legend(loc='upper left')
plt.title('Income')
plt.figure(figsize=(10, 5))
plt.stackplot(list(data), expenses_plot.values(), labels=expenses_plot.keys())
plt.legend(loc='best')
plt.title('Expenses')
plt.figure(figsize=(10, 5))
plt.stackplot(list(data), savings_plot.values(), labels=savings_plot.keys())
# plot the negative of invest on the same graph as savings
plt.plot(list(data), [-i for i in invest], label='Invested')
plt.title('Savings')
plt.legend(loc='best')
plt.show()Join Upaspro to get email for news in AI and Finance
So useful I really appreciated this!!
That is actually really handy tool. As a newbie in python, what do you suggest to be able to change the code to personalize it.
Hi Robert,
Thanks for your question. We offer a 10-session pack to get you ready to work on your own python project which requires no prior coding knowledge. We recommend you to take a look at this link.