Skip to main content

Using WASM Database for Analysis

Meta Analysis Cloud supports data analysis using a browser-based in-memory database powered by WASM technology. This allows users to analyze data files stored anywhere online, including Excel, CSV, JSON, and Parquet files.

WASM stands for WebAssembly, which is a binary format used for running high-performance code in web browsers. The goal of WebAssembly is to provide a universal web application language that can be used in parallel with JavaScript to support faster and more powerful web applications and games. Unlike JavaScript, WebAssembly can run directly in the browser without the need for an interpreter or compiler, resulting in improved performance. WebAssembly can be written in various programming languages, including C++, Rust, and Go, among others. In addition to web browsers, WebAssembly can also run in other environments such as operating system kernels, IoT devices, and servers.

Creating a WASM Model

To create a WASM model, simply click the New WASM Model button in the semantic model management interface and enter a name.

Assuming you have obtained a link to the data file you can access (e.g., https://app.mtda.cloud/assets/data/topSubscribed.csv sourced from Most Subscribed 1000 Youtube Channels), click on the "New Table" button as shown in the following image to create a table:

Click "Next" to proceed to the data preview step, where you can modify the automatically recognized field types. Click "Apply" to complete the configuration.

caution

In this file, the values of the 'Subscribers', 'Video Views', and 'Video Count' fields are stored as numbers with thousands separators. Currently, WASM DB cannot automatically recognize such numbers, so you need to change the field types to String. Alternatively, you can modify the source file to use plain number formats. You can also use the database initialization script to handle this.

Initialization Script

You can also load the data into a table using a database initialization script. Use the following SQL script: