python - How to use scikit-learn to predict the rate of change in a site's hit counts -


I want to predict the change in the hit count of a website, using the rate of change. For example, if the site hits a couple of times (5,000 to 10,000 in a day), the rate of change is 2. If the hit increases to one half (5,000 to 7,500), the rate of change is 5. The rate of sampling and change will be calculated daily. I am fine with using raw hits instead of change rate, but it seems more difficult to work with.

I want to learn science-less to predict the change in future rates. Looking at a set of previous data points, how do I predict future changes? Should I Use A Logistics Regression? A vector vector machine? anything else?

Thank you for your help! I am new to know more about science - learning, so feel free to comment.

Edit: I forgot that you were looking for scalin solution, but I think that Simple weighted moving average can be a good start. I usually try to start something simple and only if they do not give me the desired results and go to more complex things.

There are very attractive approaches, but using simple weight weighted moving average, where you give more weight for recent observations, for example: NP hit = NP. Import Array ([100, 500, 300, 800, 900]) in the form of DEF prediction (hits, weight): return np.average (hit [-lan (weight)], weight = weight) < / Code>

Result:

  gt; & Gt; Predictive (hit, [0.2, 0.3, 0.5]) 750.0 & gt; & Gt; & Gt; 900 × 0.5 + 800 * 0.3 + 300 * 0.2 750.0  

Comments

Popular posts from this blog

import - Python ImportError: No module named wmi -

Editing Python Class in Shell and SQLAlchemy -

c# - MySQL Parameterized Select Query joining tables issue -