How to Quickly Analyze Any Discrete Time-Series Data with the Right Questions

Data is never wrong, and talking to it is beautiful. The art of time-series analysis lies in asking the right questions and visualizing the answers. In this blog, I'll show you how to analyze any discrete time-series data using Canadian investment flows and CPI as examples.

Introduction

Whether you're analyzing economic indicators, financial markets, or business metrics, discrete time-series data holds valuable insights. The key is knowing which questions to ask and how to interpret the answers. Each question below unlocks a specific analytical lens, helping you develop a comprehensive understanding of your data's story.

For this tutorial, I am using data from Statistics Canada.
To generate visualizations on the fly, I am using our data viz tool, PlotsALot.

Essential Questions for Time-Series Analysis

1. "Show me a breakdown of features"

Example: "Show me a breakdown of investment flows by type"

This reveals:

• How the composition has changed over time

• Which components drive overall trends

• If certain features are more volatile than others

For Canadian investment data: This would show how direct investments, portfolio investments, and other categories have evolved relative to each other.

2. "Show me patterns across time periods/Geographic location (if applicable)"

Example: "Show me investment levels by quarter across different years"

This reveals:

• Seasonal patterns

• If certain periods consistently show higher activity

• Whether seasonality has changed over time

For Canadian investment data: This might reveal that Q4 consistently shows higher investment activity, or that seasonal patterns shifted after specific economic events.

3. "What's the ratio between key components over time?"

Example: "Show me the ratio of Canadian investments abroad to foreign investments in Canada, remove outliers"

This reveals:

• Long-term trends in relative positioning

• Periods when the relationship between components shifted

• Correlation with external events and policy changes

For Canadian investment data: This highlights when Canada became more attractive to foreign investors relative to Canadian outbound investment.

4. "How does volatility compare across time and categories?"

Example: "Compare the standard deviation of different investment types by year" or "Show me 12-month rolling averages of investments"

This reveals:

• Periods of unusual stability or instability

• Which components contribute most to overall volatility

• If volatility patterns are changing over time

For Canadian investment data: This might show that direct investments have become more stable while portfolio investments have grown more volatile.

5. "How does the data recover after disruptions? How did major events impact the data?"

Examples: "Show the trajectory of investments after the 2020 pandemic disruption" "Compare investment patterns before, during, and after the pandemic"

This reveals:

• Recovery patterns after major events

• Relative resilience of different components

• Whether the system returns to previous patterns or establishes new ones

For Canadian investment data: This would highlight which investment types recovered fastest and whether pre-pandemic patterns have resumed.

6. "What correlations exist with related indicators?"

Example: "Compare investment trends with CPI movement"

This reveals:

• Relationships between your primary data and other factors

• Potential causal connections

• Leading or lagging indicators

For Canadian investment data: This could reveal whether inflation precedes, follows, or moves independently from investment patterns.

7. "What does historical data suggest about future trends?"

Example: "Project investment trends for the next year based on historical patterns"

This reveals:

• Potential future directions

• Confidence levels for predictions

• Areas of higher uncertainty

For Canadian investment data: This would provide a data-driven forecast while highlighting limitations and assumptions.

Additional Analysis Techniques

8. Anomaly Detection

When to use: When you need to identify unusual patterns or outliers in your time-series data.

Advantages:

• Highlights data quality issues
• Identifies extraordinary events
• Reveals unexpected patterns

9. Component Decomposition

When to use: When you need to understand the fundamental structure of your time-series.

Advantages:

• Separates trend, seasonal, and residual components
• Reveals underlying patterns
• Helps in forecasting

10. Time-Lag Analysis

When to use: When you need to understand relationships between variables over time.

Advantages:

• Identifies predictive relationships
• Reveals optimal lag periods
• Helps establish causality

Conclusion

By systematically applying these analytical questions to your discrete time-series data, you can develop a comprehensive understanding that goes beyond surface-level observations. Each visualization contributes to a more complete picture of your data's behavior.