0

i like to know if i get any knowledge in R(predictive analysis) regarding Incident Management..Like if an application don't load a file from source or an alert pop up and ticket is issued for the event so on it normally get collected in a database , i assume these data may be text format or so on, which can be analysed on monthly basis ...so we will create a predictive model using any of algorithm's used.

vinosamvarghese
  • 1
  • 1
    yes you can. But for a more specific question you should specify a bit better what your desires are regarding "create a predictive model using any of algorithm's used". Your question seems like 'we have data (about incidents), what can we do with it?'. Such question is very broad. If specification is difficult then you might possibly be helped by first by going trough some introduction texts on business intelligence or flip trough examples of dashboards https://en.wikipedia.org/wiki/Dashboard_(business) for instance https://plot.ly/ – Sextus Empiricus Feb 09 '18 at 09:38
  • Thanks ... usually the incident reports will collection of text and numerical values ..so classifications and segregating is important ... in ML we can use LDA algorithms for text analysis with knn or random forest algorithms for prediction ,which will help to detect root cause of incident from the incident logs – vinosamvarghese Feb 09 '18 at 18:59
  • so your question is? – Sextus Empiricus Feb 09 '18 at 21:21
  • Have anyone has experience in creating R script for Predictive analysis of tickets(Incidents, Alerts) in IT service management etc If so kindly confirm...since i am working in a project based on it ...so i can confirm , which algorithm willbe best fitted for analysis and prediction – vinosamvarghese Feb 13 '18 at 08:45
  • "Predictive analysis of tickets", this is the title for a book not a simple question. https://www.google.com/search?q="predictive+analysis+of+tickets" – Sextus Empiricus Feb 13 '18 at 09:43

1 Answers1

1

Not sure if you are asking a question here. Share some data or how your data actally looks like.

As far as i know for predicting the incidents we first need to know how the data is distributed such as poission, exponential, linear or non-linear. If its machine data, they are usually exponential or poisson. In such cases, you can do curve fitting for prediction. If tge data is nonlinear, then Nonlinear regression, SVM or Neural networks can be used. If its linear, linear regression can be used.

By the way, if your data is time series, then time series forecasting would be another approach you could use. The functions for the above mentioned models are availabe in R.

Not_Dave
  • 206
  • 2
  • 6