Predictive Analysis
Predictive analytics refers to the application of mathematical models to large amounts of data with the aim of identifying past behavior patterns and predicting future outcomes.
Predictive analytics is a spectrum inside Data Analytics.
Practical applications of Predictive Analysis
Predictive analytics can be used to:
- Prevent credit card fraud by flagging unusual transactions
- Scoring credit card deciding whether to approve a loan or credit applications
- Predict customer churn, allowing banks to reach out right before a customer is likely to switch institutions.
Predictive Analysis quick start guide
- Understand what you can and can’t solve: it can’t replace the skills, judgment and experience of skilled professionals and you need enough data to produce an output;
- Define the most critical problems to solve;
- Identify gaps in skills and technology;
- Conduct a small pilot project;
Step-by-step guide to make a predictive analysis model
This section may be moved in the future
- Prepare historical data for statistical analysis → data need to be collected and cleansed to remove duplicate, missing, corrupted or inaccurate data, and the organized into a defined format for analysis
- Divide data into two datasets: training data and set data
- Run one or more algorithms against the dataset → decide an appropriate model type and algorithms and then run it.
Predictive modeling is an iterative process, once a learning model is built and deployed, it’s performance must be monitored and improved.
What are different types of predictive analytics models?
- Forecast models → forecast models learn from historical data to estimate values of new data
- Classification models → Sorts data togheter around common attributes.
- Hard clustering, data points either belong to a category or don’t.
- Soft clustering, data points are assigned a probability that a point belong or doesn’t belong to a category
- Outlier models → analyze abnormal entries within a dataset
- Time series modules → This model uses time as the input parameter to predict trends over a specific period.
What are the risks in predictive modeling?
- False positives
- The need of large dataset to make accurate predictions
- Bias: predictive modeling assume that the future will continue to be like the past. Thus predictive models must be refreshed with new data.