The purpose of this project is to assist the organization Alphabet Soup in analyzing which non-profit organizations are potential recipients that will effectively utilize monetary donations from Alphabet Soup. To this end, I create and optimize a neural network model for said analysis.
-
Variable IS_SUCCESSFUL is considered the target of the model.
-
Variables EIN, NAME, APPLICATION_TYPE, AFFILIATION, CLASSIFICATION, USE_CASE, ORGANIZATION, STATUS, INCOME_AMT, SPECIAL_CONSIDERATIONS, and ASK_AMT are considered the features for the model.
-
Variables that are neither targets nor features, and should therefore be removed from the input data, have not been identified.
-
8 neurons, 2 hidden layers, and activation functions ReLU and Sigmoid are selected for the model to increase accuracy of prediction.
-
I was not able to achieve the target model performance of 75% accuracy.
-
In order to increase model performance, I increase the number of nodes to 1000, 800, and 500. I also used a total of 3 hidden layers and pruned the feature USE_CASE.
Overall the attempts at optimization of model performance result in loss of 0.5806 and accuracy of 0.7266. One recommendation for solving this classification problem is to further prune the features that do not correlate with the target, which may train the model better. Changing the machine learning model may also increase accuracy by employing the appropriate algorithms to sort the data.
Data Source (file too large for upload):
- charity_data.csv
Tools:
- scikit-learn
- TensorFlow
- Pandas
Email: show.wang94@gmail.com
LinkedIn: https://www.linkedin.com/in/s-k-wang