Project detail page
A project is the main element for representing a forecasting challenge. It has an owner (a registered user in the system), and zero or more model owners (also users). A project owner can do anything to the project, including what model owners can do, but model owners are limited to creating, editing, and deleting Models, and uploading and deleting forecasts. (To become a model owner you must contact the project owner.)
Each project's detail page is divided into four vertical sections, described next: Details table, forecast models, targets, and time zeros. Currently the top looks something like this. (The project shown is a large one, so only the details table and Models sections are visible.)
Project details table
At the top of the page is a table showing non-model information related to the project:
- Summary: The same summary as on the home page: An estimate of the number of models and forecasts in the project.
- Owner: The user that has full control of the project, and includes authorization to view, edit, and delete anything (models, forecasts, etc.) The owner is the user that created the project (which is done on the home page).
- Model owners: A list of users who can view the project (applicable only if the project is private), and who can create and delete models, and upload and delete forecasts. The project owner controls specifies model owners when she creates or edits the project.
- Time Interval Type: Tells the D3 Foresight component what its x-axis label should be. Choices: Week, Biweek, or Month.
- Visibility: A project can be either public or private, which controls project visibility, including accessing data. This applies to list and detail pages, along with API endpoints. Public projects are available to anybody, users and anonymous visitors included. Private projects can only be viewed by project owners and model owners. This means that in order to grant a user access to a private account, the project owner must add her to the project's model owners list, even if that user will not be managing any models.
- Description: Prose provided by project owner. It should cover the following topics:
- Information about the "real-time-ness" of data, i.e., revised/unrevised.
- Details about how models interpret time zeros and data version dates.
- Data file naming scheme.
- Home: A link to the project's home page, which might include details about any competitions, source code, team members, included models, etc.
- Core data: A link to a page that contains data (e.g., CSV files) used to train models. For model competitions, it should include all data sets made available to everyone in the challenge, including supplemental data like Google queries or weather data. Ideally Zoltar users interested in re-running a project's models have all they need in the home and core data directories. However, they are not used directly by Zoltar.
- Truth data: Similar to the template, this is a data file that the project owner uploads. This data contains observed
target values for time zeros and locations in the project, and is a project-agnostic way to capture actual values that
models predicted for. It is used to calculate scores, and to extract the 'actual' data used in the D3 component's
TimeChart. Each Project is responsible for generating a truth table csv file in a project-specific manner. Details:
- CSV file
- No file name restrictions
Value- see FileFormats.md .
- Analysis: Links to visualization and score pages (see below for details).
The Models section lists the project's forecast models by name, with links to model detail pages (see details below). If you're a project owner or model owner then a "New" button is shown that takes you to a form for creating a model.
This section lists names of the locations in the project. These were either defined by project owner when creating the project, or were created automatically when the project template was loaded and a referenced location was not found. Here's a screen shot that shows Locations for a particular project, along with the Targets and TimeZeros sections that are described next:
This section lists information about project forecast targets. Like locations, these were either created explicitly by the project owner, or automatically when the template was loaded. The information includes the following fields.
(A note regarding auto-created targets. Auto-created targets require further, careful editing to be complete. In
particular, users must fill in the
unit fields, identify the date-related ones by checking
date, and check
Is step ahead and fill in the
Step ahead increment integer value. Importantly: You will get
unwanted results if
Is date or
Is step ahead is incorrect.)
- Name: Self-explanatory.
- Description: "". For auto-created targets, this field must be filled in.
- Unit: E.g., "percent" or "week". Unit is used when previewing data, and when downloading it. For auto-created targets, ensure that you've entered a unit. Otherwise it will be blank in previews and downloads.
- Point Value Type: todo
- Date?: This identifies whether the target is date-related or not. Examples of date-related targets might include "Season onset" or "Season peak week". For auto-created targets, the consequences of not checking off this field include forecasts being rejected as invalid if they have no point value.
- Step Ahead?: Shows two pieces of information: 1) Whether the target is a "step ahead" one, and (if so) 2) what the "step ahead increment" is. (Step ahead targets are used to predict values in the future, and are used by some analysis tools.)
- Predictions: todo
- BinLwrs: todo
This section details the project's
time zeros. About time zeros and
data version dates: Because the forecasting
field does not have standard terminology, we have settled on the following two concepts for this application. Note that
some time zeros are tagged as starting a season, specifying the season's name, which helps to segment the time zeros.
Zoltar uses season information in the visualization and score pages, where the user can select which season to show data
for. This also helps to keep performance up.
- Time zero: The date from which a forecast originates and to which targets are relative (i.e. a "2-week-ahead forecast" is two weeks ahead of the time-zero). Every forecast has a time zero.
- Data version date: An optional additional piece of metadata for a forecast. If it exists, it is associated with a time-zero. It refers to the latest date at which any data source used for the forecasts should be considered. If present, it can be used externally by system users to recreate model results by "rolling back" the core data to a particular state.
- # Forecasts: A count of the number of forecasts for this time-zero.
- Starts Season: Shows two pieces of information: 1) Whether the time zero is starts a season (it is blank if it does not start a season), and 2) the season name itself (which indicates the time zero does start a season.)