I have daily and hourly visitor data to my website (years of data) and I would like to find out if there is a pattern related to visitor traffic and calendar.
For example on which season do i have more web traffic, any relation between web traffic and days of the year (Holiday/ Work day/ week ends) , web traffic and times of the day (Morning, afternoon, evening, night).
I have also weather data which can be merged with the visitor data so i want to know if there is a pattern on certain weather conditions (ex rainy day ) and its effect on web traffic comparison .
I would appreciate if you could recommend on which methodology and a bit of instruction on how to process to process my data in KNIME.
Please note that I have already used Association rule to find out rules for the Weather conditions and traffic to certain pages on my website. It worked well because i can easily categories weather factors such as temperature in to Hot/Cold/moderate etc but i could not apply the same method to Web traffic vs weather factors because i could not categorize web traffic. in other wards i don’t know how many visitors are considered high traffic and how many visitors considered low visitor, its all relative.
Thank you for your suggestion but i thought regression is mainly used to predict the value of a numerical target while i am trying to understand weather condition such as temperature (Hot/Cold/Moderate etc values ) and its effect on web traffic comparison.
So if you are not interested in predicting visitor numbers for future days, but just want to know some general patterns, ananalyzing the correlarion between the variables could help. E.g. the Linear Correlation node could be used for this.
You could also learn a regression model with the Simple Regression Tree Learner and have a look at the learned model, since decision tree models are often well interpretable.