Anomaly detection in data
Detect anomalies and outliers in a dataset with multiple statistical methods and severity scoring.
Paste in your AI
Paste this prompt in ChatGPT, Claude or Gemini and customize the variables in brackets.
Je dois détecter des anomalies dans mon dataset [NOM_DATASET] contenant [NOMBRE_OBSERVATIONS] enregistrements et [NOMBRE_VARIABLES] variables. Les données représentent [DESCRIPTION] sur la période [PERIODE].
Type d'anomalies recherchées : [TYPE_ANOMALIES] (ex : fraudes transactions, pannes équipements, comportements clients inhabituels, erreurs de saisie)
Mets en place une détection d'anomalies complète :
- Analyse univariée : Z-score et IQR pour identifier les valeurs aberrantes par variable
- Analyse multivariée : Isolation Forest ou Local Outlier Factor pour les anomalies contextuelles
- Analyse temporelle : détection des ruptures de série et des pics inhabituels
- Règles métier : [REGLES_METIER] à implémenter comme filtres complémentaires
- Scoring des anomalies avec niveau de sévérité (critique, warning, info)
- Visualisation des anomalies détectées sur les dimensions clés
- Recommandations pour investiguer les 10 anomalies les plus critiques
Seuil d'alerte souhaité : [SEUIL_FAUX_POSITIFS] (ex : 1% de faux positifs acceptable)
Outil : [OUTIL] (Python, SQL, Excel)
Explique comment ajuster la sensibilité du système.
Personalize this prompt with Léa
Answer 3 questions and Léa tailors the prompt to your situation.
Why this prompt works
<p>The prompt is effective because it combines univariate, multivariate, and business rule methods, covering all possible types of anomalies. The severity scoring concept enables immediate prioritization of investigations.</p>
Use Cases
Expected Output
Multi-method detection system with scored anomalies, visualizations, and investigation recommendations.
Improve this prompt
Run this prompt through the Optimizer to strengthen its context, constraints and expected format.
Improve this prompt with the OptimizerComments
- LéaAI
Pour réduire les faux positifs, combinez Isolation Forest et LOF en un score moyen pondéré, puis ajustez le seuil de décision sur un échantillon labellisé. Pensez aussi à normaliser les variables avant l'analyse multivariée.
📬 Get new prompts every week
Join our newsletter and never miss a prompt.
Go further
Similar Prompts
Statistical Power Analysis
Planning a statistical experiment
Advanced Correlation Analysis
Understanding relationships between variables
Comparative Case Law Analysis
Case Law Analysis
DALL-E Prompt for Analyzing Customer Reviews
DALL-E, OpenAI's image generation tool, offers a unique visual approach to representing and communicating customer review analysis results. Rather than being limited to classic tables and charts, DALL-E allows you to create impactful visuals—infographics, illustrated sentiment clouds, or metaphorical representations—that make customer insights immediately understandable to all teams. By transforming textual data into evocative images, you facilitate decision-making and stakeholder buy-in. Whether you want to illustrate sentiment distribution, dramatize recurring pain points, or create presentation materials based on the voice of the customer, DALL-E becomes a strategic ally. This approach is particularly effective for customer experience reports, executive committee presentations, or product improvement workshops where visual impact outweighs raw numbers. Learn how to formulate your prompts to obtain faithful and actionable visualizations of your customer review data.