# Technical Data Extraction: Failed Edits per Trajectory This document contains a detailed extraction of data from two side-by-side histograms illustrating the frequency of failed edits within trajectories. ## 1. Image Overview The image consists of two histograms plotted with blue bars and black outlines. Both charts share the same X-axis label and scale but differ in their Y-axis scales and data subsets. * **Language:** English * **Chart Type:** Histogram * **Color Coding:** Blue bars (Data), Black (Axes and Text) --- ## 2. Left Chart: Failed Edits per Trajectory (All) ### Metadata and Labels * **Title:** Failed Edits per Trajectory (All) * **Y-Axis Label:** Count * **X-Axis Label:** Num. Failed Edits * **X-Axis Markers:** 0, 5, 10, 15, 20, 25, 30, 35, 40 * **Y-Axis Markers:** 0, 200, 400, 600, 800, 1000 ### Trend Analysis The data shows a **heavy-tailed distribution** (power-law-like). The vast majority of trajectories have very few failed edits (0-1), with a rapid exponential decay as the number of failed edits increases. There is a very thin "long tail" extending out to 40 failed edits. ### Data Point Extraction (Estimated) | Num. Failed Edits (Bin) | Count (Approx.) | | :--- | :--- | | 0 - 1 | ~1100 | | 1 - 2 | ~320 | | 2 - 3 | ~180 | | 3 - 4 | ~110 | | 4 - 5 | ~80 | | 5 - 10 | Decreasing from ~60 to ~20 | | 10 - 40 | Sparse distribution, mostly < 20 counts per bin | --- ## 3. Right Chart: Failed Edits per Trajectory (Resolved) ### Metadata and Labels * **Title:** Failed Edits per Trajectory (Resolved) * **Y-Axis Label:** Count * **X-Axis Label:** Num. Failed Edits * **X-Axis Markers:** 0, 5, 10, 15, 20, 25, 30, 35, 40 * **Y-Axis Markers:** 0, 25, 50, 75, 100, 125, 150, 175 ### Trend Analysis Similar to the "All" chart, this follows a **heavy-tailed distribution**. However, the total volume (Count) is significantly lower, indicated by the Y-axis maxing out at 175 rather than 1000+. This suggests that "Resolved" trajectories represent a smaller subset of the total data, but they still predominantly feature low numbers of failed edits. ### Data Point Extraction (Estimated) | Num. Failed Edits (Bin) | Count (Approx.) | | :--- | :--- | | 0 - 1 | ~170 | | 1 - 2 | ~45 | | 2 - 3 | ~22 | | 3 - 4 | ~12 | | 4 - 5 | ~8 | | 5 - 10 | Rapidly tapers to near zero | | 10 - 40 | Extremely sparse; most bins are empty or have 1-2 counts | --- ## 4. Comparative Summary * **Scale Difference:** The "All" dataset has a peak frequency roughly 6.5 times higher than the "Resolved" dataset (~1100 vs ~170). * **Consistency:** Both datasets show that the most common scenario is 0 to 1 failed edits. * **Success Correlation:** The presence of a long tail in the "Resolved" chart indicates that even trajectories with a high number of failed edits (up to ~30) can eventually reach a "Resolved" state, though this is statistically rare.

## Bar Charts: Failed Edits per Trajectory (All) and Resolved ### Overview Two side-by-side bar charts compare the distribution of failed edits per trajectory. The left chart shows all trajectories ("Failed Edits per Trajectory (All)"), while the right chart focuses on resolved trajectories ("Failed Edits per Trajectory (Resolved)"). Both charts use a y-axis for "Count" and an x-axis for "Num. Failed Edits" (0–40). The left chart has a much larger scale (up to 1000) compared to the right (up to 175). ### Components/Axes - **Left Chart (All Trajectories)**: - **Y-axis**: "Count" (0–1000, increments of 200). - **X-axis**: "Num. Failed Edits" (0–40, increments of 5). - **Bars**: Blue, decreasing in height from left to right. - **Right Chart (Resolved Trajectories)**: - **Y-axis**: "Count" (0–175, increments of 25). - **X-axis**: "Num. Failed Edits" (0–40, increments of 5). - **Bars**: Blue, similar distribution to the left chart but with lower absolute counts. ### Detailed Analysis #### Left Chart (All Trajectories) - **0 Failed Edits**: ~1000 trajectories (tallest bar). - **1–5 Failed Edits**: ~300 trajectories (next tallest bar). - **6–10 Failed Edits**: ~100 trajectories (smaller bars). - **11–40 Failed Edits**: <10 trajectories (negligible counts). #### Right Chart (Resolved Trajectories) - **0 Failed Edits**: ~175 trajectories (tallest bar). - **1–5 Failed Edits**: ~50 trajectories (next tallest bar). - **6–10 Failed Edits**: ~10 trajectories (smaller bars). - **11–40 Failed Edits**: <1 trajectory (negligible counts). ### Key Observations 1. **Dominance of Zero Failed Edits**: Both charts show a sharp drop-off after 0 failed edits, indicating most trajectories have no failures. 2. **Long Tail of Failures**: A small fraction of trajectories (e.g., 1–5 failed edits) account for the majority of non-zero failures. 3. **Resolution Impact**: The resolved trajectories chart has significantly lower counts across all failure categories compared to the "All" chart, suggesting resolution reduces failure rates. ### Interpretation The data demonstrates that **most trajectories experience no failed edits**, with a steep decline in frequency as the number of failures increases. The resolved trajectories chart reveals that even after addressing issues, a small subset of trajectories still encounters minor failures (1–5 edits), but these are far less frequent than in the unresolved dataset. This suggests that while most edits are successful, a minority of cases require iterative corrections, and resolution significantly mitigates failure rates. The long tail of failures in both charts highlights the need for targeted improvements in problematic workflows or user behaviors.