DataLila

Uncover the stories your data can tell.

Who We Are

DataLila is a service that provides a natural language interface to your data.

The state of the art in this space is advancing quite rapidly, with new papers and projects being announced almost every week! We bring the best of these advancements as a custom implementation for specific use cases relating to your data and custom schema.

Your data, or metadata, never leaves your premises! Amplify the value you derive from your data!

Offering Overview

Where We Are:
We are a team of data scientists who leverage the bleeding edge Data Science and Machine Learning techniques to help our customers derive insights from their data to help answer complex data queries.

Where We Are Headed:
Our aim is to develop an On Prem / Local solution that enables a natural language interface to your data. Get completely reproducible, explainable results and visualizations in seconds!

We begin with data cleanup.
The schema, metadata and other info is fed into our system.
A chatbot / copilot-like interface is available to ask questions about the data using a natural language
Our system will generate textual results and suitable visualizations to answer your question. But that's not all! It also generates a step-by-step explanation of the logic and code that was used to arrive at the answer.

Process

Contact us and we can help you set up an initial exploratory session.
Together we identify the scope of the project.
DataLila will staff the project, use the right tools and processes to deliver on agreed up on milestones.
We support all models of engagement from turnkey projects to IP dev and transfer.

Example Use Case

Natural Language Query Input:

“How was the performance at product level for the month of Sep'23 comparing the same with Budgets & Previous year?”

Step 1: Parse and Interpret Query

- Identify Key Terms
- ‘Performance’
- Based on knowledge of the database schema, key indicators for ‘Performance’ could include the following columns:
- Net Profit
- Gross Sales
- Net Revenue
- Volume (Only seen in the ‘Actual’ sheet)
-‘Product-Level’
- Comparisons performed in this calculation should be done after grouping the data based on each unique product
- Month/Year
- Data should be filtered on the specified month and year
- Identify suitable visualization
- Since the user is requesting a direct comparison of data from different time periods, generating grouped bar chart visualizations, one for Sep 2022 (Actual) and Sep 2023 (Actual) and one for Sep 2023 (Budget) and Sep 2023 (Actual) will best support the answer to the user’s query

Step 2: Perform Intermediate Calculations

- Filter the ‘Budget’ and ‘Actual’ tables on year==2023 and month=='Sep' then calculate the aggregate sum values for each ‘Performance’ metric and for each unique product. Record and display each result.
- Filter the ‘Actual’ table on year==2022 and month=='Sep' then repeat the same aggregate sum calculations. Record and display the result.
- Calculate the percentage change for 2022 (Actual) and 2023 (Actual) and present the resulting table to the user
- Drop the ‘Volume’ column from the 2023 (Budget) and 2023 (Actual) aggregate tables and then perform the percentage change and present the resulting table to the user

Step 3: Generate the Plot

- Using the aggregate tables generated in Step 2, create two combined dataframes, one for each comparison
- Set the index to be a MultiIndex using the Product Name and Performance Metric categories
- Using Altair, plot a grouped bar chart for each of the two combined dataframes with the following encodings:

Encoding	Value
X	Product
xOffset	Time Period
Y	Aggregate Value
Color	Time Period
Column	Metric

- Generate plot

Results

Percentage change in performance metrics from Sep 2022 (Actual) to Sep 2023 (Actual)

Product	Net Profit	Gross Sales	Net Revenue	Volume
Bliss	68.52%	47.19%	54.33%	30.23%
Cheddar Poppers	-18.99%	-8.25%	-10.75%	-19.76%
Citrus Splash	3.70%	4.36%	-17.30%	-1.8%
Lime Zest	2.39%	6.77%	1.86%	-2.36%
Munch Mix	38.93%	30.28%	46.07%	-4.13%
Nut Bar	44.82%	1.92%	28.75%	-5.16%
Seaside Bites	6.04%	12.27%	6.92%	-1.60%
Tango	30.56%	40.89%	18.66%	20.19%
Truffle	22.92%	28.15%	43.19%	11.43%

Percentage change in performance metrics from Sep 2023 (Budget) to Sep 2023 (Actual)

Product	Net Profit	Gross Sales	Net Revenue	Volume
Bliss	68.43%	32.61%	36.72%	N/A
Cheddar Poppers	6.51%	37.30%	44.66%	N/A
Citrus Splash	-6.04%	-15.31%	-28.26%	N/A
Lime Zest	-24.36%	-15.23%	-34.08%	N/A
Munch Mix	86.94%	53.55%	72.77%	N/A
Nut Bar	45.44%	27.08%	41.51%	N/A
Seaside Bites	17.07%	2.62%	0.77%	N/A
Tango	15.67%	-1.33%	-6.07%	N/A
Truffle	6.87%	0.91%	1.23%	N/A

Answer & Visualizations

Comparison Sep 2023 (Budget) vs Sep 2023 (Actual)

Based on comparisons of Net Profit for each product category, it appears that all products exceeded budget expectations except for Citrus Splash and Lime Zest in September 2023.

Comparison Sep 2022 (Budget) vs Sep 2023 (Actual)

Based on comparisons of Net Profit for each product category, it appears that all products improved their Net Profit for the company from 2022 to 2023, except for Cheddar Poppers which saw a 19% decrease in profits over that time period.

Both Comparisons

Looking at both comparisons, it looks like the Bliss product showed a nearly 70% improvement when compared to both the actual performance in 2022 and the budgeted performance for 2023. Cheddar Poppers, Citrus Splash, and Lime Zest all showed either minimal increases in performance or significant decreases, suggesting that they are currently the worst performers in the month of September. Finally, Munch Mix and Nut Bar continue to be strong performers.

DataLila
Uncover the stories your data can tell.