CoBo (Control Board) - the tool used for statistical forecasting

From Agri4castWiki
Jump to: navigation, search

What is CoBo used for?

The CoBo software used to forecast crop yield currently consists of two different tools: the CoBoNet Import Manager used to import yield, area, and production data and the CoBo Forecasts Manager used for the actual crop yield and production forecasting.

The CoBo Forecasts Manager is an integrated interface, which is used by the analysts for the forecasting, and which is used by the Statistical Administrators to manage the complete forecast analysis. As such it aims to reduce the time needed to access data, to support the forecast analysis with the automatic processing of data, and to collect and process all crop-country forecasts in order to produce coherent crop forecast tables.

CoBoNet Import Manager

This module is used by Statistical Administrators to prepare the environment for the Users, and subsequently to inspect the data. It allows to:

  • Import statistical data
  • Inspect, filter, edit, and merge the imported values
  • View and analyze the data, both in tabular and graphical form, using an interactive visualizer

For further information on how to use the CoBo Importer application, please refer to the User Guide, which you can find here: File:CoBoNet Import Manager User Guide.pdf

CoBo Forecasts Manager

Both Analysts and Supervisors can use this module to create forecasts. It allows to:

  • See and inspect existing forecasts
  • Create and manage new forecasts

For further information on how to use the CoBo Forecasts Manager application, see the User Guide: File:CoBo 3.1 Forecasts User Guide.pdf

First Step: Import and Inspection

The picture here below shows the CoBoNet Import Manager start window:

COBO Import Manager

The following summarizes the user interface elements and the tabs in the window:

  • Imports list (left-sided) - The list includes the existing imports, organized by area. You can select an import and then inspect, modify, update, and check the imported data. The most recent imports appear in bold.
  • New import button - This function allows importing new data by specifying the data sources (only enabled for Administrators).
  • Import Info tab - This tab includes the information about the import, either new or existing (the name, the area, the user, and the import date, as well as if this is the most updated import). The Import sources grid shows the data sources that compose the active import, if any.
  • Imported values tab - Once an import has been loaded or created, this tab lists all the values that have been imported. Administrators can edit these values.
  • Values checking tab - This tab allows inspecting the imported values in terms of verifying and if necessary correct the values.
  • View data tab - This tab allows viewing the data both in tabular and in graphical form. It allows saving, exporting, and managing data displays.

Second Step: Viewing and analyzing forecasts

Analysts and Supervisors use CoBo Forecasts to view, analyze, and make crop yield forecasts. They can verify how the data have been used, which analysis have been selected, and which methods have been applied. The following shows the main Control Board Console that allows selecting the Area and the Analysis:

Control Board Console

After selecting the analysis, you will be able to view the Forecasts Overview and to display the data as a graph and/or in tabular form:

Forecasts Overview

Graph View

The following summarizes how to use the other functions in the application:

  • Crop Yield Table - This function allows Analysts to compare the selected analysis with a different reference year.
  • Crop Yield Table With Analysis Comparison – This function groups the functions of the Crop Yield Table and Analysis Comparison Table, which you can compare to the selected analysis both with a different reference year and with another analysis.
  • Cereals Table – This function has the same function as the Crop Yield Table, but relates to cereals.
  • Analysis Comparison Table – This function allows comparing two analysis, that is, to compare the value of the Selected Analysis with another one. This way, the analyst can verify the difference between two different analyses.
  • Forecasts Origin Table – This function, besides showing the forecast’s value, allows verifying which method was used to create the forecast (Forecast Method).

For further information on how to use the application, see the User Guide: File:CoBo 3.1 Forecasts User Guide.pdf.

Types of Analysis

The statistical methodologies accessible through CoBo rely on the following main approaches:

Regression analysis

Regression aims to establish a relationship between a dependent variable - the yield - and one or more independent variables summarizing climate and/or weather effects on crops. Two methods are available in CoBo for regression analysis:

  • Using the CGMS Statistical Tool (CST) to calculate the regression: This tool estimates each regression coefficient (also the trend) in one-step.
  • Calculating the Regression on Residuals: Similar, but the trend is defined by the User.

For further information, please refer to the document that you can find here: File:CoBo 3.1 Log File Contents.pdf, which describes this technique in detail.

Trend Analysis

The Trend forecast represents the immediate and basic forecast within the crop yield forecasting system. A linear time trend allows representing the influence of long-term economic and technological dynamics. Besides the default linear trend, other analytical functions (such as quadratic, exponential, and logarithmic) are available, giving the analyst the ability to choose a trend that fits with the shape of the statistical yield time-series.

Scenario Analysis

Scenario analysis aims to establish similarities between past years and the present year in terms of agro-meteorological and simulated crop conditions. Once agro-climatic years are detected which are similar to the current year, the yields of these years are associated with the current season taking into account any existing yield trend.

Averages

The yield forecast can be based on the calculation of the simple or trimmed average of past years (usually, the last five years are taken into account). Normally, this method is chosen when no other method (prediction models, trend, and scenario analysis) leads to satisfactory results or when there is a lack of data - a gap - in the time series.

Last Step: Publishing Final Forecast

Once a User accesses CoBo Forecasts with Administrator rights, the analysis management functionality becomes available. After properly managing the forecast that the Analysts and Supervisors have completed, the Publish Final Forecasts functionality allows choosing which forecast to publish:

Publish Forecasts

For detailed information on how to use the Administrator functions of CoBo, please refer to the document File:CoBo Administrator User Guide.pdf