results-analysis

Name: results-analysis
Author: Boom5426

58installs28forks

How to install results-analysis?

boom5426/nature-paper-skills/results-analysis

$npx -y skills add Boom5426/Nature-Paper-Skills --skill results-analysis

Installs into the current project.

Summary

Use when analyzing experimental results, validating comparisons, generating paper-ready results text, or turning model-evaluation outputs into figures, tables, and defensible claims.

Files · 1

View on GitHub

SKILL.md

1	# Results Analysis for ML/AI Research
2
3	A systematic experimental results analysis workflow connecting experimental data to paper writing.
4
5	## Core Features
6
7	This skill provides three core capabilities:
8
9	1. Experimental Data Analysis - Read and analyze experimental data in various formats
10	2. Statistical Validation - Perform statistical significance tests and performance comparisons
11	3. Paper Content Generation - Generate text and visualizations for the Results section
12
13	## When to Use
14
15	Use this skill when you need to:
16	- Analyze experimental results (CSV, JSON, TensorBoard logs)
17	- Generate the Results section of a paper
18	- Compare performance across multiple models
19	- Perform statistical significance tests
20	- Create publication-quality visualizations
21	- Validate the reliability of experimental results
22
23	## Workflow
24
25	### Standard Analysis Pipeline
26
27	```
28	Data Loading → Data Validation → Statistical Analysis → Visualization → Writing → Quality Check
29	```
30
31	### Step 1: Data Loading and Validation
32
33	Supported Data Formats:
34	- CSV files - Tabular data
35	- JSON files - Structured results
36	- TensorBoard logs - Training curves
37	- Python pickle - Complex objects
38
39	Data Validation Checks:
40	- Completeness check - Missing values, outliers
41	- Consistency check - Data format, units
42	- Reproducibility check - Random seeds, version info
43
44	Select appropriate tools for data loading and preliminary validation based on data format.
45
46	### Step 2: Statistical Analysis
47
48	Basic Statistics:
49	- Mean
50	- Standard Deviation
51	- Standard Error
52	- Confidence Interval
53
54	Significance Tests:
55	- t-test - Two-group comparison
56	- ANOVA - Multi-group comparison
57	- Wilcoxon test - Non-parametric test
58	- Bonferroni correction - Multiple comparison correction
59
60	Select appropriate statistical tests based on data characteristics.
61
62	Key Principles:
63	- Report complete statistical information (mean ± std/SE)
64	- Specify the test method and significance level used
65	- Report p-values and effect sizes
66	- Consider multiple comparison issues
67
68	See `references/statistical-methods.md` for the complete statistical methods guide.
69
70	### Step 3: Model Performance Comparison
71
72	Comparison Dimensions:
73	- Accuracy/Performance metrics
74	- Training time/Inference speed
75	- Model complexity/Parameter count
76	- Robustness/Generalization ability
77
78	Comparison Methods:
79	- Baseline comparison - Compare with existing methods
80	- Ablation study - Validate component contributions
81	- Cross-dataset validation - Test generalization
82
83	Systematically compare performance across different methods, ensuring fair comparison.
84
85	### Step 4: Visualization
86
87	Publication-Quality Visualization Requirements:
88	- Vector format (PDF/EPS)
89	- Colorblind-friendly palette
90	- Clear labels and legends
91	- Appropriate error bars
92	- Readable in black-and-white print
93
94	Common Chart Types:
95	- Line chart - Training curves, trend analysis
96	- Bar chart - Performance comparison
97	- Box plot - Distribution display
98	- Heatmap - Correlation analysis
99	- Scatter plot - Relationship display
100
101	Use appropriate visualization tools to generate publication-quality figures.
102
103	See `references/visualization-best-practices.md` for the visualization guide.
104
105	### Step 5: Writing the Results Section
106
107	Results Section Structure:
108
109	```markdown
110	## Results
111
112	### Overview of Main Findings
113	[1-2 paragraphs summarizing core results]
114
115	### Experimental Setup
116	[Brief description of experimental configuration; details in appendix]
117
118	### Performance Comparison
119	[Comparison with baseline methods, including tables and figures]
120
121	### Ablation Study
122	[Validate contributions of each component]
123
124	### Statistical Significance
125	[Report statistical test results]
126
127	### Qualitative Analysis
128	[Case studies, visualization examples]
129	```
130
131	Writing Principles:
132	- Clearly state the hypothesis each experiment validates
133	- Guide readers to observe key phenomena: "Figure X shows..."
134	- Report complete statistical information
135	- Honestly report limitations
136
137	See `references/results-writing-guide.md` for the complete writing guide.
138
139	### Step 6: Quality Check
140
141	Checklist:
142	- [ ] All values include error bars/confidence intervals
143	- [ ] Statistical test methods are specified
144	- [ ] Figures are clear and readable (including black-and-white print)
145	- [ ] Hyperparameter search ranges are reported
146	- [ ] Computational resources are specified (GPU type, time)
147	- [ ] Random seed settings are specified
148	- [ ] Results are reproducible (code/data available)
149
150	## Common Mistakes and Pitfalls
151
152	### Statistical Errors
153
154	❌ Wrong approach:
155	- Reporting only the best results (cherry-picking)
156	- Confusing standard deviation and standard error
157	- Not reporting statistical significance
158	- Not correcting for multiple comparisons
159
160	✅ Correct approach:
161	- Report all experimental results
162	- Clearly specify

People who liked this also viewed

6 picks

Type

results-analysis

58installs28forks

View on GitHub

How to install results-analysis?

boom5426/nature-paper-skills/results-analysis

$npx -y skills add Boom5426/Nature-Paper-Skills --skill results-analysis

Installs into the current project.

Summary

Use when analyzing experimental results, validating comparisons, generating paper-ready results text, or turning model-evaluation outputs into figures, tables, and defensible claims.

Files · 1

View on GitHub

SKILL.md

1	# Results Analysis for ML/AI Research
2
3	A systematic experimental results analysis workflow connecting experimental data to paper writing.
4
5	## Core Features
6
7	This skill provides three core capabilities:
8
9	1. Experimental Data Analysis - Read and analyze experimental data in various formats
10	2. Statistical Validation - Perform statistical significance tests and performance comparisons
11	3. Paper Content Generation - Generate text and visualizations for the Results section
12
13	## When to Use
14
15	Use this skill when you need to:
16	- Analyze experimental results (CSV, JSON, TensorBoard logs)
17	- Generate the Results section of a paper
18	- Compare performance across multiple models
19	- Perform statistical significance tests
20	- Create publication-quality visualizations
21	- Validate the reliability of experimental results
22
23	## Workflow
24
25	### Standard Analysis Pipeline
26
27	```
28	Data Loading → Data Validation → Statistical Analysis → Visualization → Writing → Quality Check
29	```
30
31	### Step 1: Data Loading and Validation
32
33	Supported Data Formats:
34	- CSV files - Tabular data
35	- JSON files - Structured results
36	- TensorBoard logs - Training curves
37	- Python pickle - Complex objects
38
39	Data Validation Checks:
40	- Completeness check - Missing values, outliers
41	- Consistency check - Data format, units
42	- Reproducibility check - Random seeds, version info
43
44	Select appropriate tools for data loading and preliminary validation based on data format.
45
46	### Step 2: Statistical Analysis
47
48	Basic Statistics:
49	- Mean
50	- Standard Deviation
51	- Standard Error
52	- Confidence Interval
53
54	Significance Tests:
55	- t-test - Two-group comparison
56	- ANOVA - Multi-group comparison
57	- Wilcoxon test - Non-parametric test
58	- Bonferroni correction - Multiple comparison correction
59
60	Select appropriate statistical tests based on data characteristics.
61
62	Key Principles:
63	- Report complete statistical information (mean ± std/SE)
64	- Specify the test method and significance level used
65	- Report p-values and effect sizes
66	- Consider multiple comparison issues
67
68	See `references/statistical-methods.md` for the complete statistical methods guide.
69
70	### Step 3: Model Performance Comparison
71
72	Comparison Dimensions:
73	- Accuracy/Performance metrics
74	- Training time/Inference speed
75	- Model complexity/Parameter count
76	- Robustness/Generalization ability
77
78	Comparison Methods:
79	- Baseline comparison - Compare with existing methods
80	- Ablation study - Validate component contributions
81	- Cross-dataset validation - Test generalization
82
83	Systematically compare performance across different methods, ensuring fair comparison.
84
85	### Step 4: Visualization
86
87	Publication-Quality Visualization Requirements:
88	- Vector format (PDF/EPS)
89	- Colorblind-friendly palette
90	- Clear labels and legends
91	- Appropriate error bars
92	- Readable in black-and-white print
93
94	Common Chart Types:
95	- Line chart - Training curves, trend analysis
96	- Bar chart - Performance comparison
97	- Box plot - Distribution display
98	- Heatmap - Correlation analysis
99	- Scatter plot - Relationship display
100
101	Use appropriate visualization tools to generate publication-quality figures.
102
103	See `references/visualization-best-practices.md` for the visualization guide.
104
105	### Step 5: Writing the Results Section
106
107	Results Section Structure:
108
109	```markdown
110	## Results
111
112	### Overview of Main Findings
113	[1-2 paragraphs summarizing core results]
114
115	### Experimental Setup
116	[Brief description of experimental configuration; details in appendix]
117
118	### Performance Comparison
119	[Comparison with baseline methods, including tables and figures]
120
121	### Ablation Study
122	[Validate contributions of each component]
123
124	### Statistical Significance
125	[Report statistical test results]
126
127	### Qualitative Analysis
128	[Case studies, visualization examples]
129	```
130
131	Writing Principles:
132	- Clearly state the hypothesis each experiment validates
133	- Guide readers to observe key phenomena: "Figure X shows..."
134	- Report complete statistical information
135	- Honestly report limitations
136
137	See `references/results-writing-guide.md` for the complete writing guide.
138
139	### Step 6: Quality Check
140
141	Checklist:
142	- [ ] All values include error bars/confidence intervals
143	- [ ] Statistical test methods are specified
144	- [ ] Figures are clear and readable (including black-and-white print)
145	- [ ] Hyperparameter search ranges are reported
146	- [ ] Computational resources are specified (GPU type, time)
147	- [ ] Random seed settings are specified
148	- [ ] Results are reproducible (code/data available)
149
150	## Common Mistakes and Pitfalls
151
152	### Statistical Errors
153
154	❌ Wrong approach:
155	- Reporting only the best results (cherry-picking)
156	- Confusing standard deviation and standard error
157	- Not reporting statistical significance
158	- Not correcting for multiple comparisons
159
160	✅ Correct approach:
161	- Report all experimental results
162	- Clearly specify

People who liked this also viewed

6 picks

Type