> ## Documentation Index
> Fetch the complete documentation index at: https://docs.deepdiveplatform.com/llms.txt
> Use this file to discover all available pages before exploring further.

# ML Analyze

> Standalone ML analysis for any text content

The ML Analyze endpoint lets you run AI-powered analysis on any text, independent of social media fetching.

## Endpoint Overview

| Endpoint           | Description                 | Credits                     |
| ------------------ | --------------------------- | --------------------------- |
| `POST /ml/analyze` | Analyze text with ML models | 0.5-1 per text per analysis |

***

## Analyze Text

Run sentiment, topic, intent, and keyword analysis on any text content.

### Basic Request

```bash theme={null}
curl -X POST "https://data-api.deepdiveplatform.com/api/v1/ml/analyze" \
  -H "X-API-Key: your-api-key" \
  -H "Content-Type: application/json" \
  -d '{
    "texts": ["This product is amazing! Best purchase ever."],
    "enrich": ["sentiment"]
  }'
```

### All Analysis Types

```bash theme={null}
curl -X POST "https://data-api.deepdiveplatform.com/api/v1/ml/analyze" \
  -H "X-API-Key: your-api-key" \
  -H "Content-Type: application/json" \
  -d '{
    "texts": [
      "This product is amazing! Best purchase ever.",
      "Terrible customer service, waited 2 hours on hold.",
      "How do I return this item?"
    ],
    "enrich": ["sentiment", "topics", "intent", "keywords"],
    "categories": {
      "topic": ["product", "service", "shipping", "returns", "pricing"]
    }
  }'
```

### Parameters

| Parameter      | Type   | Required | Default          | Description                      |
| -------------- | ------ | -------- | ---------------- | -------------------------------- |
| `texts`        | array  | Yes      | -                | List of texts to analyze (1-100) |
| `enrich`       | array  | Yes      | -                | Analysis types to run            |
| `categories`   | object | No       | -                | Custom topic categories          |
| `llm_provider` | string | No       | gemini-2.0-flash | LLM provider for analysis        |

### Available Enrichments

| Type        | Description                    | Cost             |
| ----------- | ------------------------------ | ---------------- |
| `sentiment` | Positive, negative, or neutral | 0.5 credits/text |
| `topics`    | Extract main topics            | 1 credit/text    |
| `intent`    | Identify user intent           | 1 credit/text    |
| `keywords`  | Extract key phrases            | 0.5 credits/text |

### Response

```json theme={null}
{
  "success": true,
  "data": {
    "results": [
      {
        "text": "This product is amazing! Best purchase ever.",
        "enrichment": {
          "sentiment": "positive",
          "sentiment_score": 0.95,
          "topics": ["product"],
          "intent": "praise",
          "keywords": ["product", "amazing", "best", "purchase"]
        }
      },
      {
        "text": "Terrible customer service, waited 2 hours on hold.",
        "enrichment": {
          "sentiment": "negative",
          "sentiment_score": 0.12,
          "topics": ["service"],
          "intent": "complaint",
          "keywords": ["customer service", "waited", "hours", "hold"]
        }
      },
      {
        "text": "How do I return this item?",
        "enrichment": {
          "sentiment": "neutral",
          "sentiment_score": 0.50,
          "topics": ["returns"],
          "intent": "question",
          "keywords": ["return", "item"]
        }
      }
    ]
  },
  "metadata": {
    "credits_used": 9,
    "processing_time": 1.234,
    "enrichments": ["sentiment", "topics", "intent", "keywords"]
  }
}
```

***

## Custom Categories

Define your own topic categories for more relevant classification:

```bash theme={null}
curl -X POST "https://data-api.deepdiveplatform.com/api/v1/ml/analyze" \
  -H "X-API-Key: your-api-key" \
  -H "Content-Type: application/json" \
  -d '{
    "texts": ["The new iPhone camera is incredible for low-light photos"],
    "enrich": ["topics"],
    "categories": {
      "topic": ["camera", "battery", "display", "performance", "design", "price"]
    }
  }'
```

***

## Use Cases

<CardGroup cols={2}>
  <Card title="Customer Feedback" icon="comments">
    Analyze support tickets, reviews, and survey responses
  </Card>

  <Card title="Content Moderation" icon="shield">
    Detect sentiment and intent in user-generated content
  </Card>

  <Card title="Market Research" icon="chart-line">
    Extract topics and sentiment from competitor mentions
  </Card>

  <Card title="Social Listening" icon="ear-listen">
    Process imported data from any source
  </Card>
</CardGroup>

***

## Python Examples

### Batch Analysis

```python theme={null}
import requests

def analyze_texts(texts, analyses=["sentiment", "topics"]):
    response = requests.post(
        "https://data-api.deepdiveplatform.com/api/v1/ml/analyze",
        headers={"X-API-Key": "your-api-key"},
        json={
            "texts": texts,
            "enrich": analyses
        }
    )
    return response.json()

# Analyze customer feedback
feedback = [
    "Love this product!",
    "Shipping took forever",
    "Great value for money",
    "How do I get a refund?"
]

results = analyze_texts(feedback, ["sentiment", "intent"])

for result in results["data"]["results"]:
    print(f"{result['enrichment']['sentiment']}: {result['text'][:50]}...")
```

### Sentiment Distribution

```python theme={null}
from collections import Counter

def get_sentiment_distribution(texts):
    results = analyze_texts(texts, ["sentiment"])

    sentiments = Counter(
        r["enrichment"]["sentiment"]
        for r in results["data"]["results"]
    )

    total = len(texts)
    return {
        "positive": sentiments["positive"] / total * 100,
        "negative": sentiments["negative"] / total * 100,
        "neutral": sentiments["neutral"] / total * 100
    }
```

***

## Credit Calculation

Credits are calculated per text, per analysis type:

| Texts | Analyses          | Calculation              | Total      |
| ----- | ----------------- | ------------------------ | ---------- |
| 10    | sentiment         | 10 × 0.5                 | 5 credits  |
| 10    | sentiment, topics | 10 × (0.5 + 1)           | 15 credits |
| 10    | all 4 types       | 10 × (0.5 + 1 + 1 + 0.5) | 30 credits |