Strategy One

Tips and Best Practices

Best Practices When Using the Trend Analysis

  • Ensure Sufficient Data Volume for Accurate Trend Analysis

    For more accurate trend analysis results, it's important to ensure that your data volume is substantial enough. There needs to be at least three data points to run a trend analysis.

  • Use High-Quality Continuous Time-Based Data for Trend Analysis

    While Strategy performs lightweight data processing before trend analysis, such as filling in missing metric data, it's advisable to conduct trend analysis on continuous, high-quality time-based data. Trend analysis results can be compromised if there's a significant amount of missing metric data. It's important to note that trend analysis may fail or yield suboptimal outcomes if attributes contain NULL or NaN (Not-a-Number) values. While the algorithm fills in missing metric values, this may lead to an inaccurate trend analysis.

  • Statistical Significance

    When conducting time series linear trend analysis, it's crucial to consider the statistical significance of the observed trend. Statistical significance provides an objective measure of whether the observed trend is likely due to an actual underlying pattern or merely a result of random fluctuations in the data. Use the tooltip to access critical metrics like R, R-Squared, and p-value.

Optimize Trend Analysis Line Chart Interpretation

Get the most out of your trend analysis line chart visual!

  • Info icon Hover over the info icon to see information about the statistical properties of the underlying model and its parameters.

  • Expand for Clarity If you need more detailed information and display the entire set of data points, expand the visualization or set Container Fit to None. This enhances the readability of the chart.

Trend Analysis FAQs

What time series trend analysis method is used in the current release

The time series trend analysis method used in the latest release is called linear regression. This method is described on the OTexts web site. You can refer to the statsmodel Python module (Time Series analysis tsa - statsmodels 0.14.0 ) for more technical details.

What types of attributes and metrics are required for successful trend analysis?

To successfully plot trend for your data, you need either:

  • Single attribute of type date or datetime + single metric of interest (recommended)

  • Single attribute of type int + metric of interest (when date/datetime attribute is unavailable for time series trend analysis)

Is missing data handled in trend analysis?

Yes, the algorithm can automatically fill in missing values for a metric if it has corresponding time-series data. This is only done for the calculation, so the result visualization will still show missing data. The algorithm uses forward and backward fill, using the neighboring value, depending on the direction. The primary method is forward grab, which takes the previous neighbor. If the first value is missing, backward fill is run to estimate a value.

What time-based attribute trend analysis levels are supported?

Time-based attribute trend analysis is supported for Date, DateTime, or integer-based time attributes. As long as an attribute is one of the above types, the algorithm supports all time levels.