# Outlier Detection Engine MCP for AI Agents AI Agent Connect > Outlier Detection Engine identifies statistical anomalies in large datasets using deterministic Z-Score and IQR methods. It stops your AI client from guessing which rows are outliers by using actual math. Use it to clean data, find sensor errors, or scrub datasets for machine learning without hitting context limits. ## Overview - **Category:** artificial-intelligence - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_dfkaEjEHL0xoHc706KKIKrxSgf6YRwqXhyMOeE7z/ai-agent-connect - **Tags:** statistical-analysis, anomaly-detection, z-score, iqr, data-cleaning, math-engine ## Description When you're dealing with thousands of rows of data, asking an AI to find the weird stuff is a recipe for disaster. Most models will start guessing based on a few examples they see in the context window, often missing real anomalies or flagging perfectly normal data just because it looks different. This Connector changes that by handing the heavy lifting off to a dedicated math engine. Instead of the AI trying to feel its way through your spreadsheet, it uses the Outlier Detection Engine to run actual calculations. You get precise results based on standard deviations and quartiles. It's like giving your agent a calculator that actually knows statistics. You can find these tools in the Vinkius marketplace, where they're organized to make connecting these types of data-heavy capabilities to your workflow a lot faster. It's about moving from vibe-based data cleaning to deterministic results that you can actually trust for your reports or machine learning models. ## Tools ### detect_outliers Run a Z-Score or IQR check on a specific column to find data points that fall outside the normal range. This tool provides the exact math needed to identify statistical anomalies. ## Prompt Examples **Prompt:** ``` Find all rows where the 'Temperature' reading is a statistical outlier using Z-Score > 3. ``` **Response:** ``` Found 4 outliers. The most extreme is row 142 with Temperature 98.5°C (Z-Score = 4.1), followed by row 87 (Z = 3.8), row 201 (Z = 3.4), and row 15 (Z = 3.1). ``` **Prompt:** ``` Check the 'Price' column for anomalies using the robust IQR method with a 1.5 multiplier. ``` **Response:** ``` Using an IQR threshold of 1.5, I identified 12 items priced significantly above the upper bound of $450 (Q3 + 1.5×IQR). These appear to be luxury or premium-tier products. ``` **Prompt:** ``` Are there any abnormal network latency values in this monitoring dataset? ``` **Response:** ``` Yes. Using Z-Score analysis, 3 network requests had ping times exceeding 3 standard deviations (Z > 3): rows 44 (Z=3.9), 128 (Z=3.5), and 302 (Z=3.2). ``` ## Capabilities ### Calculate exact Z-Scores The Connector computes the standard deviation from the mean for every row in your dataset. ### Identify IQR anomalies It finds data points that fall outside the interquartile range for robust detection. ### Set custom sensitivity You can define exactly how many standard deviations or what multiplier counts as an outlier. ### Scan thousands of rows The engine processes large amounts of data locally on your machine in seconds. ### Flag specific data points Your agent gets a list of exact row numbers that failed the statistical test for review. ## Use Cases ### Sensor Malfunctions An engineer has 50,000 rows of temperature data. They use the tool to find spikes that indicate a failing sensor in a manufacturing plant. ### Fraud Detection A financial analyst scans thousands of transactions to find spending amounts that are statistically impossible for a specific user profile. ### Data Cleaning A researcher has a messy CSV of survey results and needs to strip out troll entries that don't follow the standard response pattern. ### Quality Control A factory manager checks production speeds to identify specific machines that are underperforming compared to the fleet average. ## Benefits - Stop LLM hallucinations by replacing intuition with deterministic Z-Score and IQR calculations. - Handle massive datasets that would otherwise exceed your AI client's context window by processing data locally. - Get exact boundary values for every flagged row so you know exactly why a data point was marked as an outlier. - Choose between Z-Score for normal distributions and IQR for skewed data to match your specific dataset's shape. - Set your own sensitivity levels, like a 3.0 standard deviation, to control exactly how aggressive the detection is. ## How It Works The bottom line is that your agent stops guessing and starts using math to find data errors. 1. Provide your agent with a dataset and specify the column you want to check for errors. 2. Tell the agent which math method to use, such as Z-Score for normal data or IQR for skewed data. 3. Receive a list of specific row numbers and their calculated values that fall outside your chosen bounds. ## Frequently Asked Questions **How does the Outlier Detection Engine find errors in my data?** It uses standard statistical methods like Z-Score and IQR to calculate how far each data point sits from the average. If a point is far enough away, the engine flags it for you. **Can I use the Outlier Detection Engine on very large CSV files?** Yes. Because the math happens locally on your machine, you can scan thousands of rows without worrying about your AI client's memory limits or context window. **What is the difference between Z-Score and IQR in this Connector?** Z-Score is best for data that follows a normal bell curve. IQR is a more robust method that works better for skewed data, like income or real estate prices. **Will the Outlier Detection Engine help me clean my data for machine learning?** Absolutely. It helps you identify and remove 'poison' data points that could skew your models and lead to inaccurate predictions. **How do I set the sensitivity for the Outlier Detection Engine?** You can tell your agent exactly what threshold to use, such as a Z-Score higher than 3 or a specific multiplier for the IQR method. **Can the Outlier Detection Engine handle skewed data like house prices?** Yes, by using the IQR method. This is the preferred way to find anomalies in data that isn't perfectly balanced or normally distributed. **What is the difference between Z-Score and IQR?** Z-Score assumes data is normally distributed and is sensitive to extreme outliers. IQR is based on percentiles (25th and 75th), making it robust and ideal for skewed or non-normal data. **Can I customize the outlier sensitivity threshold?** Yes! You set the threshold parameter: typically 3 for Z-Score (flagging values beyond 3 standard deviations) or 1.5 for IQR (the standard Tukey fence multiplier). **Does it automatically remove the outliers?** No. The engine flags the outliers and provides their exact Z-Scores or IQR bounds so the AI can report them to you. The decision to drop or keep them remains with you.