I am working on a binary classification task with a pretty straightforward input set of numeric features. One of these features is particularly good, but it cannot be used in real life because it's a measure that is obtained after the fact has occurred. Is it possible to predict this measure based on the other features, and then build a model including this predicted measure?
In more detail, I am building a classifier for this challenge from the UCI repo: https://archive.ics.uci.edu/ml/datasets/bank+marketing
The feature that cannot be used is the call duration because one can't know how long a call will last before it takes place. So I am wondering, could I build a regression model or at least a binned classifier to predict how long a call will last before it takes place, then feed this prediction to the model and replace the provided call duration feature?
Is this approach, something that is acceptable? has it been done before? or am I just inventing something new that has never been done before?
It is possible to build a regression model or a binned classifier to predict the duration of a call before it takes place. However, the accuracy of the model will depend on the quality and availability of the data used to train the model.
A regression model would be appropriate if you are trying to predict a continuous variable, such as the duration of a call-in seconds or minutes. You can use different techniques like linear regression, decision tree, Random Forest, etc. to train the model. You would need to have a dataset of past calls that includes the duration of each call and any relevant information that could be used to predict the duration of future calls, such as the time of day, the caller's location, and the purpose of the call.
Alternatively, you could use a binned classifier model to predict call duration, which would classify calls into different duration categories (e.g., short, medium, long) instead of predicting a specific duration. This type of model would also require a dataset of past calls and relevant information to train the model.
If you have a good dataset and the model is trained correctly it could be possible to replace the provided call duration feature with the prediction from the model, However, it's important to keep in mind that the accuracy of the model will depend on the quality of the data used to train the model and the complexity of the model. Also, it would be important to test the model on new data to evaluate its performance, and make sure that it generalizes well.
Hatem Ali Other characteristics may be used to predict call duration, and this projected call duration can then be used as a feature in your binary classification model. This method, known as feature engineering, can frequently result in enhanced model performance. However, it is critical to recognize the various restrictions and biases that this technique may entail, as well as properly analyze the performance of the final model. It is also worth mentioning that this strategy has already been used and is a common technique in Machine Learning.
- Badges
- Science topic
- Similar topics
- Biological Science
- Biochemistry
- Biomolecules
- Peptides