Item: IBM Watson Studio on Cloud Pak for Data
Rating: 10
Author: John Robert Uy

Overall Satisfaction with IBM Cloud Starter Kits

Use Cases and Deployment Scope

Used to test prototype applications for clients. Mostly used for creating predictive data models, descriptive models, and basic ETL. There are plans to test new speech-to-text and image-to-text applications for new clients in 2021. Cloud storage is a secondary use since it is the only platform that supports older or legacy databases

Pros and Cons

Pros

Auto AI makes creating predictive models so much easier and faster. It creates several models and ranks them according to precision (or accuracy) allowing us to rapidly select the most optimized model. While the models are not perfect at the first run, it gives us an idea on which models to focus on cutting the turnaround times from 3 days to less than 4 hours.
The cloud structure allows us to reuse datasets that are in different projects. This cuts down the need to create new pipelines or ETL steps.

Cons

Auto AI allows us to select the best models to use when creating predictive models. The app ranks and lists down the models according to accuracy (or precision0. This alone is worth the subscription as it cut down our turnaround times from 3 days to 1 day.

Return on Investment

Positive: reduce our turnaround times were reduced from 3 days to 1 day. This allows us to create more models and service more clients.

Alternatives Considered

Google BigQuery and Google Cloud AI

Easy to use, but still requires a lot of coding to use. There is no ranking of models used and models are not persistent, which means you have to keep running the models again every time you leave the session. The filesystem is clunky and need to keep authorizing Google Drive to save any datasets.

Key Insights

Do you think IBM Watson Studio on Cloud Pak for Data delivers good value for the price?

Yes

Are you happy with IBM Watson Studio on Cloud Pak for Data's feature set?

Yes

Did IBM Watson Studio on Cloud Pak for Data live up to sales and marketing promises?

Yes

Did implementation of IBM Watson Studio on Cloud Pak for Data go as expected?

Yes

Would you buy IBM Watson Studio on Cloud Pak for Data again?

Yes

Other Software Used

KNIME Analytics Platform, Weka.IO, Google Cloud Storage

Likelihood to Recommend

It is well suited for mechanistic models such as time series and descriptive analysis. There is almost no code necessary when creating these models, as it takes the guesswork out of setting up the parameters. There are also models that we've never before even tried because of its complexities but AutoAI shows that these models are sometimes best for the given problem.

Comments

Please log in to join the conversation

Auto AI is a must have for every Data Analyst