# November 17 2022

## Datasets

### New Design of Dataset Analysis Page

Our dataset details page has a new look and feel

![](https://slabstatic.com/prod/uploads/j51jqwst/posts/images/5hdbRSpwXJn-W8eNKfzyCHlY.gif)

### Data Profiling support for text and categorical datasets

You can now register your text or categorical datasets and view detailed profiling information.

![](https://slabstatic.com/prod/uploads/j51jqwst/posts/images/SU1bkS6K2cEypu5Y-4TdgKsz.gif)

**Data summary:**

Details about the features, missing or duplicate values

**Insights:**

We now compute insights about features with "similar distributions", "skewness", and "negative values" and more about different features in your datasets.

**Variable Details**

* Summary of different features in your dataset
* Frequency charts, word cloud, pie charts, letter details for **categorical variables**
* Quantile statistics, KDE Plot, QQ Plot, and Box plot for **numeric variables**

**Correlation**

For your registered datasets, you could see Pearson Correlation, Spearman Correlation, and KendallTau Correlation.

**Missing Value Chart**

Missing values in your dataset can be easily visualized through a frequency bar chart.

### Topic Modeling Analysis

For text-based datasets, we now support probabilistic topic modeling based on LDA/ HDP.

![](https://slabstatic.com/prod/uploads/j51jqwst/posts/images/jen3PVbyBCYpegkbKFuLxS0u.png)

## Notebook

### Support for Markov Python Package

MarkovML's Python Package is now supported inside Jupyterhub.

Simply `import markov` to get started.

![](https://slabstatic.com/prod/uploads/j51jqwst/posts/images/eKEv3fX-GNvLw4xEH7kmXeQS.png)

### Dataprep Library

MarkovML's Jupyter Notebook now has`dataprep` a library pre-installed. You can profile. You can quickly run an EDA on datasets

```python
from dataprep.datasets import load_dataset
from dataprep.eda import create_report
df = load_dataset("titanic")
create_report(df).show()
```

## Other Enhancements

### Revamped Onboarding Experience

We have improved our onboarding experience. After signing up, users are shown some ways to use MarkovML.

![](https://slabstatic.com/prod/uploads/j51jqwst/posts/images/oIR22L0FLMvB1Jx8n3uZz1um.gif)

### Switch between multiple workspaces

Users who are part of multiple workspaces can now easily switch between them.

Enterprise users can easily use MarkovML for work and personal use cases. Data between each workspace is stored separately.

![Users who are part of multiple workspaces, can switch their workspace by pressing switch button on the top header bar](https://slabstatic.com/prod/uploads/j51jqwst/posts/images/p13V3uIfAFI35ige_QUQHeS6.gif)

![markov switch allows users to change their active workspace while working with MarkovML's python package](https://slabstatic.com/prod/uploads/j51jqwst/posts/images/HChNXFQ6hhNGGBDDniYvrPGo.gif)

### Bug fixes and stability improvements