

I have data in two columns, X and Y where the function Y changes for increasing values of the X variable. If Y shows variations with X, the simplest way to do this is by fitting the data with a linear function.(In python), but I am looking for some "advanced" strategy (other than linear regression, i.e., using Neural Networks or other machine learning techniques) to elaborate on these data because i also want to find pattern in it. Need guidance .
modified 9Aug23 13:21pm.





Stop reposting the same question all over the site. ONCE in QA is enough.





To be fair, Andre did tell him to repost it here.
"These people looked deep within my soul and assigned me a number based on the order in which I joined."
 Homer





Thanks, I made the suggestion as it is an AI related question.





I wasn't aware that Q&A was "nonAI" related. It's a poor "AI question" to start with, and the whole concept of "moving the question" has made this painful.
"Before entering on an understanding, I have meditated for a long time, and have foreseen what might happen. It is not genius which reveals to me suddenly, secretly, what I have to say or to do in a circumstance unexpected by other people; it is reflection, it is meditation."  Napoleon I





Hi Gerry, I followed suit as the more experienced posters tend to send al AI questions to the AI forum because, like I do, we tend to think that a better answer or response will be achieved there than in the Q&A.
I will refrain from making the suggestion again.





The redirects are usually to people who post questions relating to the CodeProject AI, and we point them to the CodeProject.AI Discussions[^] forum, rather than the general AI one here. And yes, I agree it can be confusing.





This is not remotely in my field of expertise, I decided to ask AI about an AI question. Please check and confirm each suggestion as it is commonly known that false values are given by ChatGPT in certain scenarios. Suggestions given by ChatGPT 
You can explore more advanced techniques like polynomial regression, decision trees, random forests, support vector machines, and neural networks.
Polynomial Regression: This involves fitting a polynomial equation to the data, allowing for curved relationships between X and Y. You can experiment with different polynomial degrees to capture varying degrees of complexity.
Decision Trees and Random Forests: Decision trees split the data based on different conditions and can handle both linear and nonlinear relationships. Random forests are an ensemble of decision trees that can handle complex relationships more effectively.
Support Vector Machines (SVM): SVMs are good for both linear and nonlinear patterns. They aim to find a hyperplane that best separates the data while considering a "kernel trick" to map the data into a higherdimensional space for better separation.
Neural Networks: Neural networks are capable of capturing complex patterns and relationships in data. You can design a network with multiple layers and nodes to model intricate nonlinear connections between X and Y.
Gradient Boosting: This ensemble technique combines multiple weak learners (usually decision trees) to create a strong predictive model. It's powerful for capturing complex relationships.
KNearest Neighbors (KNN): This instancebased learning algorithm uses nearby data points to predict the value of a new point. It can capture local patterns that may not be apparent with global models.
Clustering Algorithms: If you're interested in discovering inherent patterns in your data without a specific Y variable, clustering techniques like kmeans or hierarchical clustering might be useful.
Feature Engineering: Transform your features (X variables) to create new ones that might capture complex relationships more effectively. This might involve interactions, logarithmic transformations, or other functions.
Time Series Analysis: If your data has a temporal component, time series techniques like ARIMA (AutoRegressive Integrated Moving Average) or LSTM (Long ShortTerm Memory) networks could be useful.
Dimensionality Reduction: Techniques like Principal Component Analysis (PCA) or tSNE (tDistributed Stochastic Neighbor Embedding) can help you visualize and analyze highdimensional data.
Remember that selecting the best technique depends on your specific data and problem. Experiment with different methods, possibly even combining some, to see which one works best for uncovering the patterns you're interested in. Also, don't forget to split your data into training and testing sets to evaluate the performance of each technique accurately.
End of ChatGPT, I hope this points you in the right direction.





great answer, thanks !
when can i enroll in your classes ?
cheers, Bill
«The mind is not a vessel to be filled but a fire to be kindled» Plutarch





Thanks Bill, had a good chuckle just now.





I'm curious if anyone on these forums or project are familiar with or have attended / published / coauthored at either USENIX or OSDI.
I attended this week, and there's some absolutely staggering things happening in the ML/Ai world beyond what's being reported, especially in both hardware and simple scope.
Some of the papers and case studies absolutely floored me. But others showed that there's a lot of open source advancements that aren't paywalled out there (unless you consider minor things like access to Nvidia H100s and such a pay wall haha)





? LP = not(true(LP)).
LP = not(true(LP)).
? unify_with_occurs_check(LP, not(true(LP))).
false.
Here is the answer from ChatGPT
Please explain this in English:
? LP = not(true(LP)).
LP = not(true(LP)).
? unify_with_occurs_check(LP, not(true(LP))).
false.
ChatGPT
The first statement in Prolog is an assignment where LP is defined as the negation of the truth value of LP itself. This means that LP is the logical negation of LP, indicating that LP is false if it is true, and true if it is false.
The second statement is a query that checks if there exists a unification between LP and not(true(LP)). It attempts to find a consistent value for LP that satisfies the equation.
In this case, the result of the query is false, indicating that there is no valid unification between LP and not(true(LP)). In other words, there is no consistent value that can simultaneously satisfy the equation LP = not(true(LP)). This suggests that there is a contradiction in the logic, as the equation cannot hold true for any value of LP.
=======================================
It turns out that ChatGPT is incorrect because Prolog is merely recognizing the pathological selfreference of LP. "true" and "not" in the above context are meaningless placeholders to Prolog.
Prolog equally rejects this expression: X = foo(bar(X)).
I was trying to show that Prolog recognizes and rejects the Liar Paradox. It turns out the Prolog recognizes and rejects every expression that is isomorphic to the Liar Paradox.
modified 16May23 17:59pm.






To find an answer.
1. Find an AI client
2. Figure out how to use it.
3. Ask it.






Don't expect serious answers here. Rather, consult a dictionary.
If this is a homework handin problem, be particularly careful about directly copying the answers you get here.
It is probably better to ask ChatGPT





It is the art of preventing your employees from getting on with their work undisturbed.





Stopping short of achieving your level of incompetence.
"Before entering on an understanding, I have meditated for a long time, and have foreseen what might happen. It is not genius which reveals to me suddenly, secretly, what I have to say or to do in a circumstance unexpected by other people; it is reflection, it is meditation."  Napoleon I





Doing it successfully means that one does not get angry when one sees silly questions.







i want to make a machine capable of detecting and classifying passes and throwins based on football match video. could you help me?





Has anyone else had success running CodeProject AI in a Docker container? I've spent days trying to troubleshoot what is going on with my deployment without success.
Here's my setup: At the base I'm running ESXi. Blue Iris is running in a Win10 VM. For the Docker setup, I'm running PhotonOS in a VM, with Portainer on top to give me a GUI for Docker. Inside Docker, I'm pulling in the codeproject/aiserver image. I'm using macvlan as the networking config to give it an IP on the LAN.
The CPAI container spins up fine, I can access the web GUI fine. I can ping it from the BI server. But it doesn't respond to any kind of requests. I've noticed that if I go into the CPAI Explorer, none of the models are showing up except for a few that appear to be defaults. If I load up an image in Explorer and ask it to analyze it, nothing happens. I just get a timeout and no logs generated.
I've triple checked that I've got the model folder mapped to "/app/modules/ObjectDetectionYolo/custommodels". I've ssh'd to the container and verified the models show up in there, but CPAI doesn't seem to recognize them.




