no_time_to_name

This program is about a Machine learning algorithm which looks at consumer activity and predicts his/her spending for a time period. The implementation of this enables additional functionality.

ORIGINAL IDEA: Based on the customer’s spending history/pattern, the program implements an algorithm to predict the customer's spending and hence allocate a realistic set amount of money each month, based on the customer's wishes, that they can spend. For example: If a customer spends more in March and December, he/she will get more spendable money in their account compared to the other months. Customer gets an alert if payment exceeds the expected amount and gets an additional notification to disable credit card if that is the case. If someone spends less money than they are expected to, the remaining money (from the Checking 1) goes to a rewards account with higher interest rates (promotes saving). If a customer overspends the excess amount spent will incur a higher interest rate (again promotes saving). If a customer is allowed access to this data, he/she can plan out her spendings and savings more evenly.

The monthly prediction algorithm can also be used to predict if a person is eligible to take a loan and how long they might take to repay the loan. It may be implemented alongside credit scores to deem if a person is credit-worthy.

#Challenges: The major challenge was to implement a machine-learning algorithm which can both handle large amounts of data and produce reliable results. We had tried up to 13 different algorithms before we set on a certain one. Moreover, another major challenge was to define the functions and their dependencies. The scikit-learn module does not allow integration of the traditional formulation with machine learning and so we had to define meaningful functions. Ultimately, the challenge which overcame us was TIME and we simply couldn't finish the CLI to interface the other classes with.

Built With

• python
• tons-of-modules-i-wont-name-no-time