' Matrix coding query for theme co-occurrence
Query.MatrixCoding(
  Rows: Nodes["Themes\AI Benefits"],
  Columns: Nodes["Themes\Challenges"],
  Scope: Documents["Interviews\*"]
)

Code "AI Adoption" AND Code "Resistance"
WITHIN QUOTATION DISTANCE 50

Themes
├── AI Benefits
│   ├── Efficiency
│   └── Accuracy
└── Challenges
    ├── Technical barriers
    └── User resistance

Required:
  - analysis_method: "Statistical analysis to perform (quantitative or qualitative)"
  - language: "R, Python, SPSS, Stata, NVivo, ATLAS.ti, MAXQDA"
  - variable_info: "Variable names, types (for quant) OR coding scheme (for qual)"

Optional:
  - data_file: "File path/format"
  - special_requirements: "APA format, Korean support, inter-rater reliability, etc."
  - caqdas_version: "Software version for compatibility"
  - coding_method: "Thematic analysis, content analysis, grounded theory"

## Analysis Code

### Analysis Information
- **Analysis Method**: [Method name]
- **Language**: [R/Python/SPSS/Stata]
- **Required Packages**: [Package list]

### 1. Environment Setup

```r
# ============================================
# [Analysis Name] Analysis Script
# Created: [Date]
# R version: 4.x.x
# ============================================

# Set seed for reproducibility
set.seed(2024)

# Install and load required packages
if (!require("pacman")) install.packages("pacman")
pacman::p_load(
  tidyverse,   # Data processing
  car,         # Assumption checking
  effectsize,  # Effect size
  ggpubr       # Visualization
)

# Load data
data <- read_csv("data.csv")

# Check data
glimpse(data)

# Check missing values
sum(is.na(data))

# Convert variable types (if needed)
data <- data %>%
  mutate(
    group = factor(group),
    gender = factor(gender)
  )

# Descriptive statistics by group
data %>%
  group_by(group) %>%
  summarise(
    n = n(),
    mean = mean(score, na.rm = TRUE),
    sd = sd(score, na.rm = TRUE),
    se = sd / sqrt(n),
    ci_lower = mean - 1.96 * se,
    ci_upper = mean + 1.96 * se
  )

# Normality test
shapiro.test(data$score[data$group == "A"])
shapiro.test(data$score[data$group == "B"])

# Q-Q plot
qqPlot(data$score, main = "Q-Q Plot")

# Homogeneity of variance test
leveneTest(score ~ group, data = data)

# Run [analysis method]
result <- [analysis_function]

# Summary of results
summary(result)

# Calculate effect size
effect <- cohens_d(score ~ group, data = data)
print(effect)

# Multiple comparisons (for ANOVA)
TukeyHSD(result)

# Results graph
ggplot(data, aes(x = group, y = score, fill = group)) +
  geom_boxplot(alpha = 0.7) +
  geom_jitter(width = 0.2, alpha = 0.5) +
  stat_summary(fun = mean, geom = "point",
               shape = 18, size = 4, color = "red") +
  labs(
    title = "[Analysis Results]",
    x = "Group",
    y = "Score"
  ) +
  theme_pubr() +
  theme(legend.position = "none")

ggsave("results_plot.png", width = 8, height = 6, dpi = 300)

# APA format results
# "[Analysis method] results showed [statistic] was statistically
# [significant/not significant], [statistic = X.XX, p = .XXX,
# effect size = X.XX, 95% CI [X.XX, X.XX]]."

## Prompt Template

## Code Template Library

### Independent t-test (R)
```r
# Independent samples t-test
t_result <- t.test(dv ~ iv, data = data, var.equal = TRUE)
# Welch's t-test (when equal variance assumption violated)
t_result <- t.test(dv ~ iv, data = data, var.equal = FALSE)
# Effect size
cohens_d(dv ~ iv, data = data)

# One-way ANOVA
aov_result <- aov(dv ~ iv, data = data)
summary(aov_result)
# Effect size
eta_squared(aov_result)
# Post-hoc test
TukeyHSD(aov_result)

# Multiple regression
lm_result <- lm(dv ~ iv1 + iv2 + iv3, data = data)
summary(lm_result)
# Check multicollinearity
vif(lm_result)
# Standardized coefficients
lm.beta(lm_result)

# Mediation analysis (process package)
library(processR)
process(data = data, y = "dv", x = "iv", m = "mediator",
        model = 4, boot = 5000)

# ============================================
# Thematic Analysis with R
# Packages: tm, quanteda, tidytext
# ============================================

library(tidyverse)
library(tm)
library(quanteda)
library(tidytext)

# Load qualitative data
interviews <- read_csv("interviews.csv")

# Create corpus
corpus <- corpus(interviews, text_field = "transcript")

# Tokenize
tokens <- tokens(corpus,
                remove_punct = TRUE,
                remove_numbers = TRUE,
                remove_symbols = TRUE)

# Create document-feature matrix
dfm <- dfm(tokens) %>%
  dfm_remove(stopwords("english"))

# Word frequency analysis
word_freq <- textstat_frequency(dfm, n = 50)

# Visualize top words
ggplot(word_freq[1:20,], aes(x = reorder(feature, frequency), y = frequency)) +
  geom_col() +
  coord_flip() +
  labs(title = "Top 20 Most Frequent Words",
       x = "Word", y = "Frequency") +
  theme_minimal()

# Co-occurrence network
fcm <- fcm(dfm, context = "window", window = 5)
textplot_network(fcm, min_freq = 10, edge_alpha = 0.5)

# ============================================
# Thematic Analysis with Python
# Packages: nltk, spacy, textblob
# ============================================

import pandas as pd
import nltk
from nltk.corpus import stopwords
from collections import Counter
import spacy
from textblob import TextBlob
import matplotlib.pyplot as plt
from wordcloud import WordCloud

# Load spaCy model
nlp = spacy.load("en_core_web_sm")

# Load data
df = pd.read_csv("interviews.csv")

# Preprocess text
stop_words = set(stopwords.words('english'))

def preprocess(text):
    doc = nlp(text.lower())
    tokens = [token.lemma_ for token in doc
             if not token.is_stop and not token.is_punct]
    return tokens

df['tokens'] = df['transcript'].apply(preprocess)

# Word frequency
all_words = [word for tokens in df['tokens'] for word in tokens]
word_freq = Counter(all_words)
top_20 = word_freq.most_common(20)

# Visualize
words, counts = zip(*top_20)
plt.figure(figsize=(12, 6))
plt.barh(words, counts)
plt.xlabel('Frequency')
plt.title('Top 20 Most Frequent Words')
plt.tight_layout()
plt.show()

# Word cloud
wordcloud = WordCloud(width=800, height=400,
                     background_color='white').generate(' '.join(all_words))
plt.figure(figsize=(15, 8))
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis('off')
plt.show()

# ============================================
# Content Analysis: Inter-Rater Reliability
# Package: irr
# ============================================

library(irr)

# Load coding data (2 raters)
# Format: Each row = document, each column = rater's code
coding_data <- read_csv("coding_ratings.csv")

# Cohen's Kappa (2 raters, nominal categories)
kappa_result <- kappa2(coding_data[, c("rater1", "rater2")])
print(kappa_result)
# Interpretation:
# < 0.00: Poor agreement
# 0.00-0.20: Slight agreement
# 0.21-0.40: Fair agreement
# 0.41-0.60: Moderate agreement
# 0.61-0.80: Substantial agreement
# 0.81-1.00: Almost perfect agreement

# Krippendorff's Alpha (multiple raters, any measurement level)
# Data format: rows = raters, columns = documents
kripp_result <- kripp.alpha(t(coding_data[, -1]), method = "nominal")
print(kripp_result)

# Percent Agreement
agree(coding_data[, c("rater1", "rater2")])

# ============================================
# Content Analysis: Inter-Rater Reliability
# Package: sklearn.metrics
# ============================================

import pandas as pd
from sklearn.metrics import cohen_kappa_score, confusion_matrix
import numpy as np

# Load coding data
df = pd.read_csv("coding_ratings.csv")

# Cohen's Kappa (2 raters)
kappa = cohen_kappa_score(df['rater1'], df['rater2'])
print(f"Cohen's Kappa: {kappa:.3f}")

# Percent Agreement
agreement = (df['rater1'] == df['rater2']).sum() / len(df)
print(f"Percent Agreement: {agreement:.1%}")

# Confusion matrix
cm = confusion_matrix(df['rater1'], df['rater2'])
print("Confusion Matrix:")
print(cm)

# Krippendorff's Alpha (requires simpledorff package)
# pip install simpledorff
import simpledorff

# Format: rows = documents, columns = raters
reliability_data = df[['rater1', 'rater2']].values.T
alpha = simpledorff.calculate_krippendorffs_alpha_for_df(
    pd.DataFrame(reliability_data)
)
print(f"Krippendorff's Alpha: {alpha:.3f}")

# ============================================
# NVivo Project Setup Automation
# Requires: NVivo API or manual import
# ============================================

import pandas as pd

# Generate NVivo classification sheet
nvivo_setup = {
    'Name': ['Interview_01', 'Interview_02', 'Interview_03'],
    'Type': ['Interview', 'Interview', 'Interview'],
    'Participant_ID': ['P001', 'P002', 'P003'],
    'Gender': ['Female', 'Male', 'Female'],
    'Years_Experience': [5, 10, 3]
}

df = pd.DataFrame(nvivo_setup)
df.to_excel('nvivo_classification_import.xlsx', index=False)

# Generate auto-coding rules template
autocoding_rules = """
# NVivo Auto-Coding Rules

## Rule 1: Keyword-based coding
Node: Themes\\AI Benefits
Keywords: benefit, advantage, positive, improve, enhance
Context: NEAR (within 10 words)

## Rule 2: Pattern-based coding
Node: Themes\\Challenges
Pattern: difficult*, challenge*, barrier*, obstacle*

## Rule 3: Sentiment-based coding
Node: Sentiment\\Positive
Criteria: Sentences containing: love, great, excellent, wonderful
Exclude: NOT, but, however (within 5 words)
"""

with open('nvivo_autocoding_rules.txt', 'w') as f:
    f.write(autocoding_rules)

print("NVivo project setup files created!")

# ============================================
# ATLAS.ti Code-Document Frequency Table
# ============================================

import pandas as pd

# Simulated coding data
# Format: document_id, code, quotation_count
atlas_data = {
    'Document': ['Interview_01', 'Interview_01', 'Interview_02',
                'Interview_02', 'Interview_03', 'Interview_03'],
    'Code': ['AI_Benefits', 'Challenges', 'AI_Benefits',
            'User_Resistance', 'Efficiency', 'Challenges'],
    'Count': [5, 3, 7, 2, 4, 6]
}

df = pd.DataFrame(atlas_data)

# Pivot to code-document matrix
code_doc_matrix = df.pivot_table(
    values='Count',
    index='Document',
    columns='Code',
    fill_value=0
)

print("Code-Document Frequency Matrix:")
print(code_doc_matrix)

# Export for ATLAS.ti import
code_doc_matrix.to_excel('atlas_code_document_matrix.xlsx')

# Co-occurrence calculation
from itertools import combinations

codes_per_doc = df.groupby('Document')['Code'].apply(list)
cooccurrence = {}

for doc, codes in codes_per_doc.items():
    for code1, code2 in combinations(set(codes), 2):
        pair = tuple(sorted([code1, code2]))
        cooccurrence[pair] = cooccurrence.get(pair, 0) + 1

print("\nCode Co-occurrence:")
for pair, count in sorted(cooccurrence.items(), key=lambda x: -x[1]):
    print(f"{pair[0]} <-> {pair[1]}: {count}")