Credit Risk Dataset Analysis

I'll start by examining the data source and loading the relevant skills.

Now let me load the slide workflow reference for the PPT generation:

Excellent! I now have a clear picture of the dataset and the workflow. Let me start with Phase 1 — exploration to understand the data.

The dataset is a credit risk dataset with 35 fields covering demographics (age, gender, education, occupation, marital status), financial attributes (income, credit score, loan amount, loan term), and risk indicators (default status, delinquencies, credit cards, existing loans).

Let me explore the data

Interesting — the default_count is 0, meaning all records are "No Default". Let me explore the data more deeply to understand the risk profile distribution, income, loan amounts, credit scores, and demographic segments.

Now I have a thorough understanding of the data. Let me proceed to Phase 2 — generating charts, then the PPT report.

Key findings summary:

• 1,000 records, all with "No Default" status (no defaults recorded in this dataset)
• Average credit score: 579 (Fair range), average loan: $27,259, avg income: $28,825
• 35% of customers are in Poor credit range (300-499), another 16% in Fair (500-579)
• Credit score is the strongest differentiator of risk profile
• Occupation groups are fairly evenly distributed; White Collar and Professional each ~21%
• Age 45-54 is the largest segment (21.4%), with highest average loan amounts
• Delinquencies: 61.6% have 0 delinquencies, 30.6% have 1
• Correlation: Loan amount is most correlated with credit cards (0.18) and loan term (0.14)

Let me now generate the charts:

I need to use the exact member short names from the schema. Let me fix the aliases:

All 6 charts generated successfully. Now let me also create a couple more charts for the report — a pie chart for credit score segments and a bar chart for work years analysis

Now let me write the summary and outline, then proceed with the PPT pipeline. First, let me select a template

Now let me proceed with the template selection

Great choice! Let me use the "pulse" template and proceed with the guideline

Guideline is done with 10 pages mapped and all 10 charts recognized. Now let me generate the presentation

Let me load the dashboard skill to build the interactive BI dashboard

Now let me load the design template, references, and explore the data source

Now let me get the explore details and design the dashboard. First, let me get the explore info and check what charts we already have from the analysis

Now I have all the information I need. Let me write the dashboard design doc and then create the charts and dashboard. I'll reuse the existing analysis charts and create new ones for the KPIs. Let me first check what embed charts are already on disk, then create the new ones needed.

Now I need to create the 4 KPI big_number charts. Let me load the embed-chart skill for chart JSON reference, then create the charts

Let me read the sample chart files for big_number and cartesian to understand the exact JSON structure

Now I have all the reference material. I need to query the actual data to get the KPI values for the big_number charts, then write the chart JSON files. Since the analysis charts already exist from save_embed_chart, I can reuse them. Let me create the KPI charts and the dashboard JSON.

First, let me get the KPI values:

Now let me create the 4 KPI charts and the dashboard JSON. I'll use save_embed_chart for the KPI big_number charts since they work with Cube data:

Now let me check what chart files actually exist on disk to use the correct filenames

All 14 charts are on disk. Now let me also verify which chart JSON files exist as chart_*_config.json files for the dashboard to reference correctly

Let me check the UUID of the existing chart files to get the correct field IDs and tableName

Perfect. Now I have all the information needed. Let me also check the KPI chart JSON structure to understand the big_number format

Now I have all the field IDs. Let me write the dashboard JSON:

Let me fix the markdown tile height