How can I create a new Dataset
There are 4 ways to create datasets:
- Uploading by hand from a CSV or XLS.
- Automatically connecting it to a file in Google Drive
- As a result of running a pipeline (read more here)
- Using the API or Python library (read more here)
For the first two, on the Create new button at the top right, click on Dataset, after which you can drag the CSV or XLSX file you want to upload or select the Google Drive file you want to connect.
How to connect your data
Make sure your CSV, XLSX or file from Google Drive has a column with a Date. If you choose the first option, simply select the file from your computer. For the second option, make sure to sign in with Google in order to synchronize your data from Google Drive. Once you have given Alphacast permission, you will be able to add a file from your Google Drive. Select the file you desire. With this option, updates will be made automatically. If you choose an XLSX or CSV, when data is added, you will have to upload your file again.
How to configure the columns of a dataset?
Once the file is selected, you must choose to configure the dataset, defining the Entity columns and the variables. The Entity columns are those necessary to uniquely define a row of the dataset (to see more about entities click here) and all datasets must have at least one “Entity” column with the date (to see the formats of accepted date click here) Then you define the type of each variable (if it is a text or a number), and indicate which variables you want to ignore. Finally, the last step allows you to choose the name of the dataset and its repository.
What is an “Entity”?
The Entity columns are necessary to uniquely define a row of the dataset, the combinations of Entities cannot be repeated. For example, if your only entity is Date your dataset cannot have repeating dates. If your entities are Date and Country the dates can be repeated, but the combinations of date and country must be unique.
What are the accepted formats for date columns when creating a dataset?
When creating a new dataset, the date column must be selected, marked as entity and, in data type, select the format. YY-MM-DD (year, month, day) is generally used, but can be changed with the Change Date Format button.
For instance, in the following case the format is YYYY-MM-DD and you can click on Change Date Format to match it with the format you have on your file, like it is shown in the image. Also, you can see the Python date formats for reference.
Naming and publishing a Dataset
The last tab Name your dataset allows you to choose the name of your dataset and the repository where it will be stored.