Application Insights Stats-2

August 16, 2015

Report Abuse
This model consumes dataset produced by Application Insights Stats-1 experiment and produces various rolling window statistics.
<h1>Application Insights Stats-2</h1> <h3>Summary</h3> This model consumes dataset produced by Application Insights Stats-1 experiment and produces various rolling window statistics for Page View Performances data of Application Insights. Azure Application Insights enable <a href="https://azure.microsoft.com/en-in/documentation/articles/app-insights-code-sample-export-telemetry-sql-database/#start-continuous-export-to-azure-storage">Continuous Data Export</a> that can be consumed for further use. <h3>Input Dataset</h3> Data source consumed in this experiment (i.e. available as .CSV file) is comprised of following items: <br><li>Date <br><li>Pagename <br><li>Country <br><li>Metric_Name <br><li>Min.by.Day <br><li>Max.by.Day <br><li>Mean.by.Day <br><li>Median.by.Day <br><li>Standard Deviation.by.Day <br><li>Percentile_10.by.Day <br><li>Percentile_90.by.Day <h3>Experiment Flow</h3> This experiment performs the following steps to produce the required statistics: <br><br><strong>Step-1:</strong> Read the source data (Output Dataset-1 produced by App-Insight Stats-1) <br><br><strong>Step-2:</strong> Define metadata for the given dataset, i.e. <br><li><strong>group_by_cols</strong>: Number of columns to be used for grouping. These columns must be starting columns in the data set <br><li><strong>metric_cols</strong>: Number of metric columns, these columns will follow group_by_cols in the data <br><li><strong>fn_list</strong>: List of functions to be executed on each of the metric column <br><li><strong>fn_args</strong>: Arguments corresponding to the functions in the fn_list <br><li><strong>stats_col_names</strong>: Name of each column produced by executing various function on the metric data <br><li><strong>summary_by_name</strong>: Name for various summary periods (i.e. Week, Month, Quarter, Year) <br><li><strong>summary_by_period</strong>: Number of days for each summary period (i.e. 7, 30, 90, 365) <br><br><strong>Step-3</strong> Loop through each of the metric column and function list. With this each function is executed on every metric column. Data is collated to produce the output result dataset. <h3> Output Dataset-1</h3> Summarized statistics produced by various summary periods (i.e. Week, Month, Quarter, Year), Page name, Country and Metric name. The results are stored in .CSV file having following items: <br><li>Date <br><li>Pagename <br><li>Country <br><li>Metric_Name <br><li>Min.by.Week, Min.by.Month, Min.by.Quarter, Min.by.Year <br><li>Max.by.Week, Max.by.Month, Max.by.Quarter, Max.by.Year <br><li>Mean.by.Week, Mean.by.Month, Mean.by.Quarter, Mean.by.Year <br><li>Median.by.Week, Median.by.Month, Median.by.Quarter, Median.by.Year <br><li>SD.by.Week, SD.by.Month, SD.by.Quarter, SD.by.Year <br><li>Percentile_10.by.Week, Percentile_10.by.Month, Percentile_10.by.Quarter, Percentile_10.by.Year <br><li>Percentile_90.by.Week, Percentile_90.by.Month, Percentile_90.by.Quarter, Percentile_90.by.Year <h3>Output Dataset-2</h3> It is produced by normalizing Output Dataset-1 for consumption in any reports