How to create a forecast  
with IBM Cognos Analytics

With IBM Cognos Analytics, you can forecast tabular datasets based on regression analysis to identify likely scenarios for how the data will evolve.

Why should companies create forecasts with IBM Cognos Analytics?

Forecasts based on regression analyzes are important wherever data is displayed over time: It is therefore particularly worthwhile in the field of online marketing to predict various metrics using regression analysis with statistical probability . These can be various metrics, such as the development of clicks, impressions, conversion rates, cost per click (CPC), cost per conversion (CPA), ROAS or cost per deal from advertising budget .

With a reliable forecast – based on inventory data – the most likely developments can be extrapolated months in advance, such as downward or upward trends . This in turn helps the marketing team of the company in question to adapt strategies and resource planning in order to prevent negative scenarios from occurring in the first place . In short: The company has an information advantage, can take preventive action and has stable early detection with forecasts from regression analyses . With IBM Cognos Analytics you can create such forecasts for your data from online marketing and for various metrics.

Create a forecast using regression analysis with IBM Cognos Analytics from a tabular data set.

How does forecasting work with IBM Cognos Analytics?

The creation of forecasts using regression analyzes is possible with different analysis environments, such as via Microsoft Power BI for Desktop or with a specially trained machine learning model with Vertex AI on the Google Cloud . But this is also possible with IBM Cognos Analytics: Here you can upload tabular data sets in CSV format or connect BigQuery databases and cloud storages to analyze the data. Once you have configured your data set and found the right chart visualization, you can use the forecasting function of IBM Cognos Analytics.

The most important thing here is the use of seasonality values ​​in IBM Cognos Analytics : A seasonality indicates the causal development pattern for the data forecast in the forecast and is measured in the data points of the underlying tabular data set . Each table row below the header shows the entries for the time dimension on a time axis (either in days, hours, minutes, seconds, months, calendar weeks or years), as well as the corresponding metrics that occurred at these times in other columns (e.g : clicks, impressions, conversion, deals, cost per conversion, etc.). So if we have two columns in the table, with oneTime axis in the left column (360 days in 360 rows) and clicks from organic Google search in the right column (clicks on said days in 360 rows) and we enter a seasonality of 360 days , the forecast shows the development of the entire period used as a development pattern for regression analysis.

Forecasting with IBM Cognos Analytics: Using seasonality data for a forecast with regression analysis.


Create forecast with different visualizations in IBM Cognos Analytics

You can use IBM Cognos Analytics to create a forecast with different visualizations. For technical reasons, the forecast function is not possible for every diagram visualization. Line charts or bubble charts, for example, are suitable for forecasts . For analysis purposes other than forecasts, there are also tables, bullet charts, area charts, box plots or heat maps.

Create a forecast with IBM Cognos Analytics: In the IBM Cognos Analytics visualization gallery, you can select the appropriate chart for your forecast. Line charts and bubble charts are good for forecasting with regression analysis.


Upload tabular data sets or connect external data like a cloud storage

Before a forecast can be created using IBM Cognos Analytics, a tabular data set with a timeline in one column must be uploaded. It is important that the time axis (e.g. dates of all days of a year, hours, months, calendar weeks, years) should be complete . So if dates of all days in a certain period are used as a time axis in a column, all days must be listed completely and without gaps in the rows of the column. Unless you want to work with tabular data sets for your forecast in IBM Cognos Analytics, you can also connect ERP systems, CRM systems, BigQuery databases or cloud storage to create a forecast from the inventory data.

Prepare your data for forecasting with IBM Cognos Analytics.

Use IBM Cognos Analytics to create a forecast

You can use IBM Cognos Analytics for your forecasts if you are already using the free 30-day trial. The full version of IBM Cognos Analytics starts at $10 a month.

Example forecast on a timeline with IBM Cognos Analytics based on regression analysis with seasonality values.

Slava Wagner

Marketing & Lead Gen

SEA, paid media, conversion rate optimization, market and trend analysis in the Berlin-Brandenburg area.

Telephone: +49 176 588 744 04

Free consultation: Create forecasts with IBM Cognos Analytics

If you don’t know how to create a forecast with IBM Cognos Analytics, or which seasonality values ​​you should use for the regression analysis, then feel free to ask your question in a free consultation:

Ask your question

Do you have a question regarding IBM Cognos Analytics?

    FAQ - Summary: Create forecasts with IBM Cognos Analytics

    Here you will find a summary of important and frequently asked questions about creating forecasts with IBM Cognos Analytics:

    First, open the IBM Cognos Analytics dashboard, upload your tabular dataset with timeline in CSV format or connect a BigQuery database, then choose a chart type from the IBM Cognos Analytics visualization gallery. You can then use the forecast function for line charts or bubble charts with a time axis. Click on the chart to see the forecast button. Enter your seasonality value for the regression analysis here, as well as the forecast period and activate the forecast.

    A seasonality in a forecast using regression analysis in IBM Cognos Analytics describes the data period that indicates a causal development pattern for a forecast. Seasonality is represented as a metric in the form of whole numbers, and each counter corresponds to a data point on the timeline of your inventory data. So, in your tabular CSV dataset, if you enter 200 consecutive days in the rows under the date column, and choose a forecast seasonality of 200, then the dynamics of the entire period of your dataset will be reflected in the forecast.

    You can use a wide variety of external data sources for a forecast with IBM Cognos Analytics, such as BigQuery databases. But you can also upload tabular data in the form of a CSV table and create a forecast. Importantly, you can only create a forecast with IBM Cognos Analytics for data with a timeline. This means that in the uploaded tabular dataset there must be a time column, under which a time axis is broken down in the rows below, such as consecutive with date in each row. This can also be years, months, calendar weeks, hours, minutes or seconds.

    A forecast period reflects the number of data points of the timeline specification in a tabular CSV data set in a forecast. If you have 200 consecutive days in 200 rows as a timeline in a date column, and you choose a forecast period in the forecast with IBM Cognos Analytics of 200, the forecast is based on the period for the next 200 days from the last day of the tabular data set.

    Overview: Create a forecast with IBM Cognos Analytics

    With the business intelligence application IBM Cognos Analytics you can create forecasts based on inventory data of a tabular CSV dataset. IBM Cognos Analytics can import and combine data from various sources, including external databases, tabular datasets, cloud storages and big data platforms. It is important that your data basis contains a timeline, on the basis of which you can create a forest.

    In the IBM Cognos Analytics visualization gallery, you can select the appropriate line or bubble chart for your forecast and then use the forecast function. Here you can store the seasonality value that indicates the causal development pattern for the regression analysis.

    With forecasts for metrics in online marketing, you can identify upwards and downwards trends 6 – 12 months in advance and thus preventively anticipate negative scenarios before they occur and expand positive scenarios in order to get even more potential out of the synergy effects.