Data
This page describes the data to upload and how to upload it
Data To Upload
The first three data sets in the table below are used to filter the recommendations. The filter data is typically uploaded every 24 hours. The final data set - the Transaction History - is a one time only upload and is a initial feed to get good recommendations from day one. All data sets are optional, but are recommended to get the most out of Avail SaaS.
Name Description Specification Sample
Category Data Which category does each product belong to. A product can belong to zero, one or several categories. The maximum number of categories is 32767. Used to filter predictions. Example: If a user is browsing products in category A, you can choose to only recommend product within category A. Column 1 - ProductId
Column 2 - CategoryId
X1021cctt27
X10217uiop
q6231fret34
AaR21cctt27
......
Category Names Assigns names to category identifiers. Used for a more friendly administration of categories in the Control Panel. Column 1 - CategoryId
Column 2 - Category Name
cctt27Drama
7uiopScience Fiction
fret34Action
......
Valid Products Data on which products are currently "on the shelves". Used as a master filter to set which products are recommendable. Column 1 - ProductId
X1021
q6231
AaR21
...
Product Data Product attribute data. The purpose of the product data is to create rules. The different product data types are described below. Inform Avail about the names and data types of the columns you are going to upload. An arbitrary number of columns
Column 1 - ProductId
Column 2 - Product Data
...
Column N - Product Data
X1021Gone in 60 seconds16.95gone.jpg
q6231The Killing Fields11.95fields.jpg
AaR21Matrix19.95matrix1.jpg
............
Transaction History Order data. The transaction history is used to derive the predictions. Each row is an order row, i.e. a purchase made by a customer. Only upload real products and users. Do not upload for example test users, proxy users such as call centers, give-aways, shipment-and-handling, vat and so forth. As only distinct rows are taken into consideration you can save bandwidth by making sure that only distinct rows are uploaded. If the third optional column cart id is used, cart predictions will be available. Distinct Rows
Column 1 - UserId
Column 2 - ProductId
Optional:Column 3 - OrderId
abc123X1021
abc123q6231
abc123AAR2144
qeewde42144
qeewq6234
......
or
abc123X1021fe55
abc123q6231c444
abc123AAR2144c444
qeewde42144ee23
qeewq6234ee23
.........
Identifiers
Allowed characters in identifiers are the alphanumeric characters and underscore, at sign, full stop and hyphen-minus. The alphanumeric character set consists of the numbers 0 to 9 and letters A to Z in the English alphabet case-sensitive. The maximum length is 500 characters
Name Description
CategoryId A unique identifier for a category
OrderId A unique identifier for an order placed
ProductId A unique identifier for a product
UserId A unique identifier for a user
Category Names
Allowed values are arbitrary characters. Reserved values are the null value 'null' and the crypto prefix '_enc_'. The maximum length is 500 characters.
Product Data Specification
Name Format Min Max Example
Date yyyy-MM-dd 1915-01-01 2036-12-31 2008-12-05
Float A decimal number. Dot as a delimiter. See IEEE 754-1985 for formal definition. Easy definition but slightly incorrect: -16,777,216 with 6 decimals accuracy. Formally: -3.4028235 × 10^38 Easy definition but slightly incorrect: 16,777,216 with 6 decimals accuracy. Formally: 3.4028235 × 10^38 10.95
Int A natural number. I.e. whole numbers including zero and negative numbers. -2,147,483,647 2,147,483,647 110
String A string of arbitrary length consisting of UTF8 characters. Empty String 20,000 characters image.jpg, Star Wars.
Uploading Data
Upload data using Avail's Data Exporter which can be found under the tab Resources. You specify a set of SQL queries, stored procedures/functions or file names and the data is uploaded automatically to Avail SaaS. You can upload data anytime except Tuesdays 8:00-12:00 CET which is reserved for maintenance slots.

Proprietary data exports are possible but not recommended. For instructions on how to do this see here.