We have the data set with variables describing the natural phenomena for the long time period. The variable are for each hour of the day.
First task, is to select the month for the most occurrence of the phenomena.
Then, base on the steps outlined below, to develop the logistic regression model.
The most important, to display the steps how to develop this model, how to evaluate this model.
How to evaluate the relationship between the phenomena occurrence frequency and probability derived from the logistic regression.
• Using the logistic regression built in SAS software to develop a phenomena occurrence simulation model.
• Evaluate the model performance using the measurements of logistic regression.
• Evaluate the model performance by analyzing 1) the relationships between a phenomena occurrence frequency and probability derived from the logistic regression and 2) the relationships between the daily number of hours of phenomena occurrences and probability derived from the logistic regression. These evaluations can be displayed by tables or figures.
• Prepare a 1-2 page report on your approach. This report should explain how it works, and why it get the results like this.
Prepare a 3-5 page report on the [url removed, login to view] part will be completed later. But if this possible to discuss,would be great.
The data set is in the table, in SAS format. Variable, which is dependent, the predictant, represents the every day when event occur as 1, and every day without event as 0 (dichotomic). The model should produce the forecast from 24 hours ahead to 5 days ahead.
I need this done ( some model runs) and report first draft by Monday, May 21st.