When to build separate models
$begingroup$
I'm pretty new to predictive modeling, but am interested in generating predictions for credit card account spend. These are existing accounts.
The data I have available to me is Card Type (i.e. Platinum, Black, Gold) and spend/transaction data over the last year.
I have a two questions:
How do I decide whether to use linear regression or a decision tree/random forest? Or do I try many techniques?
I expect spend will differ considerably across Card Types; does this suggest I should build separate models by Card Types? If I had a another set of attributes that I thought might segment customers, but wasn't sure, how would I evaluate that?
If there are guides online that answer the above, that'd be much appreciated as well. Haven't found any good ones (or at least I think not)
predictive-modeling regression
asked Dec 7 '18 at 4:56
pmodeler22pmodeler22
61
$endgroup$
bumped to the homepage by Community♦ 4 hours ago
This question has answers that may be good or bad; the system has marked it active so that they can be reviewed.
add a comment |
$begingroup$
I'm pretty new to predictive modeling, but am interested in generating predictions for credit card account spend. These are existing accounts.
The data I have available to me is Card Type (i.e. Platinum, Black, Gold) and spend/transaction data over the last year.
I have a two questions:
How do I decide whether to use linear regression or a decision tree/random forest? Or do I try many techniques?
I expect spend will differ considerably across Card Types; does this suggest I should build separate models by Card Types? If I had a another set of attributes that I thought might segment customers, but wasn't sure, how would I evaluate that?
If there are guides online that answer the above, that'd be much appreciated as well. Haven't found any good ones (or at least I think not)
predictive-modeling regression
asked Dec 7 '18 at 4:56
pmodeler22pmodeler22
61
$endgroup$
bumped to the homepage by Community♦ 4 hours ago
This question has answers that may be good or bad; the system has marked it active so that they can be reviewed.
$begingroup$
Usually you don't build separate models unless you really cannot separate cases well. The model depends on your data and your objective, when in doubt try many and see what works better.
$endgroup$
– user2974951
Dec 7 '18 at 7:57
add a comment |
$begingroup$
I'm pretty new to predictive modeling, but am interested in generating predictions for credit card account spend. These are existing accounts.
The data I have available to me is Card Type (i.e. Platinum, Black, Gold) and spend/transaction data over the last year.
I have a two questions:
How do I decide whether to use linear regression or a decision tree/random forest? Or do I try many techniques?
I expect spend will differ considerably across Card Types; does this suggest I should build separate models by Card Types? If I had a another set of attributes that I thought might segment customers, but wasn't sure, how would I evaluate that?
If there are guides online that answer the above, that'd be much appreciated as well. Haven't found any good ones (or at least I think not)
predictive-modeling regression
asked Dec 7 '18 at 4:56
pmodeler22pmodeler22
61
$endgroup$
I'm pretty new to predictive modeling, but am interested in generating predictions for credit card account spend. These are existing accounts.
The data I have available to me is Card Type (i.e. Platinum, Black, Gold) and spend/transaction data over the last year.
I have a two questions:
How do I decide whether to use linear regression or a decision tree/random forest? Or do I try many techniques?
I expect spend will differ considerably across Card Types; does this suggest I should build separate models by Card Types? If I had a another set of attributes that I thought might segment customers, but wasn't sure, how would I evaluate that?
If there are guides online that answer the above, that'd be much appreciated as well. Haven't found any good ones (or at least I think not)
predictive-modeling regression
predictive-modeling regression
asked Dec 7 '18 at 4:56
pmodeler22pmodeler22
61
asked Dec 7 '18 at 4:56
pmodeler22pmodeler22
61
asked Dec 7 '18 at 4:56
pmodeler22pmodeler22
61
asked Dec 7 '18 at 4:56
pmodeler22pmodeler22
61
asked Dec 7 '18 at 4:56
pmodeler22pmodeler22
61
61
bumped to the homepage by Community♦ 4 hours ago
This question has answers that may be good or bad; the system has marked it active so that they can be reviewed.
bumped to the homepage by Community♦ 4 hours ago
This question has answers that may be good or bad; the system has marked it active so that they can be reviewed.
$begingroup$
Usually you don't build separate models unless you really cannot separate cases well. The model depends on your data and your objective, when in doubt try many and see what works better.
$endgroup$
– user2974951
Dec 7 '18 at 7:57
add a comment |
$begingroup$
Usually you don't build separate models unless you really cannot separate cases well. The model depends on your data and your objective, when in doubt try many and see what works better.
$endgroup$
– user2974951
Dec 7 '18 at 7:57
$begingroup$
Usually you don't build separate models unless you really cannot separate cases well. The model depends on your data and your objective, when in doubt try many and see what works better.
$endgroup$
– user2974951
Dec 7 '18 at 7:57
$begingroup$
Usually you don't build separate models unless you really cannot separate cases well. The model depends on your data and your objective, when in doubt try many and see what works better.
$endgroup$
– user2974951
Dec 7 '18 at 7:57
add a comment |
2 Answers
2
active
oldest
votes
$begingroup$
short answer: assess simple models first, and then build more complex ones if necessary.
For a roadmap, I'd go like this:
- get familiar with the different categories of machine learning problems: classification vs regression problems. Your problem falls in the regression category (at least at first sight, you could also reframe it into a classification problem if necessary)
- get familiar with the main corresponding algorithms
- get familiar with the strategies to train / test a model performance
- since you are explicitly asking for online resources, you can have a look at Andrew Ng's machine learning course (especially the first weeks of the course). It's not free, but you can also have a general overview buy looking for his Machine Learning Yearning Book. Another resource coming to my mind is Machine Learning Mastery website (can't remember the guy's name) where you can find some introductory ebooks to answer those questions. It's well written and can provide you a good lift
Does this help?
answered Dec 7 '18 at 13:20
seisemanseiseman
1092
$endgroup$
$begingroup$
I'm not a fan of this answer. You are continuing with the assumption that its OK to immediately go into modeling. The data has to be studied first.
$endgroup$
– I_Play_With_Data
Jan 9 at 13:59
add a comment |
$begingroup$
You are committing some classic mistakes that people make when first entering the world of data science. Your question should not be, "What model do I use?". The question should be, "What can I learn from my data?"
First mistake you're making is jumping straight to modeling - never do that. You need to go through a period of exploratory data analytics (EDA) to help you understand your data. The whole point of an EDA phase is to enable you to ask smarter questions of your project and when you come up with those questions, you will be in a much better position to determine what models you will need. Also, keep in mind that EDA can also serve to answer analytical questions. In your example, your question of spend by card type can probably be answered during a thorough EDA phase.
Second mistake you're making is thinking about the type of models you need without thinking about the questions you want to answer. What's the business case you're trying to solve? What's your hypothesis? What data will you have available when you will run predictions in the future? Do you have a supervised on unsupervised problem on your hands? Etc, etc, etc. These are all things that you need to have some answers to before you can even think about the models that will be used.
Don't forget, this is data science. You need to approach your problems in a methodical, scientific manner in order to truly achieve the results that you seek.
answered Jan 9 at 13:54
I_Play_With_DataI_Play_With_Data
880419
$endgroup$
add a comment |
Your Answer
StackExchange.ifUsing("editor", function () {
return StackExchange.using("mathjaxEditing", function () {
StackExchange.MarkdownEditor.creationCallbacks.add(function (editor, postfix) {
StackExchange.mathjaxEditing.prepareWmdForMathJax(editor, postfix, [["$", "$"], ["\\(","\\)"]]);
});
});
}, "mathjax-editing");
StackExchange.ready(function() {
var channelOptions = {
tags: "".split(" "),
id: "557"
};
initTagRenderer("".split(" "), "".split(" "), channelOptions);
StackExchange.using("externalEditor", function() {
// Have to fire editor after snippets, if snippets enabled
if (StackExchange.settings.snippets.snippetsEnabled) {
StackExchange.using("snippets", function() {
createEditor();
});
}
else {
createEditor();
}
});
function createEditor() {
StackExchange.prepareEditor({
heartbeatType: 'answer',
autoActivateHeartbeat: false,
convertImagesToLinks: false,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: null,
bindNavPrevention: true,
postfix: "",
imageUploader: {
brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
allowUrls: true
},
onDemand: true,
discardSelector: ".discard-answer"
,immediatelyShowMarkdownHelp:true
});
}
});
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fdatascience.stackexchange.com%2fquestions%2f42266%2fwhen-to-build-separate-models%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
2 Answers
2
active
oldest
votes
2 Answers
2
active
oldest
votes
active
oldest
votes
active
oldest
votes
$begingroup$
short answer: assess simple models first, and then build more complex ones if necessary.
For a roadmap, I'd go like this:
- get familiar with the different categories of machine learning problems: classification vs regression problems. Your problem falls in the regression category (at least at first sight, you could also reframe it into a classification problem if necessary)
- get familiar with the main corresponding algorithms
- get familiar with the strategies to train / test a model performance
- since you are explicitly asking for online resources, you can have a look at Andrew Ng's machine learning course (especially the first weeks of the course). It's not free, but you can also have a general overview buy looking for his Machine Learning Yearning Book. Another resource coming to my mind is Machine Learning Mastery website (can't remember the guy's name) where you can find some introductory ebooks to answer those questions. It's well written and can provide you a good lift
Does this help?
answered Dec 7 '18 at 13:20
seisemanseiseman
1092
$endgroup$
$begingroup$
I'm not a fan of this answer. You are continuing with the assumption that its OK to immediately go into modeling. The data has to be studied first.
$endgroup$
– I_Play_With_Data
Jan 9 at 13:59
add a comment |
$begingroup$
short answer: assess simple models first, and then build more complex ones if necessary.
For a roadmap, I'd go like this:
- get familiar with the different categories of machine learning problems: classification vs regression problems. Your problem falls in the regression category (at least at first sight, you could also reframe it into a classification problem if necessary)
- get familiar with the main corresponding algorithms
- get familiar with the strategies to train / test a model performance
- since you are explicitly asking for online resources, you can have a look at Andrew Ng's machine learning course (especially the first weeks of the course). It's not free, but you can also have a general overview buy looking for his Machine Learning Yearning Book. Another resource coming to my mind is Machine Learning Mastery website (can't remember the guy's name) where you can find some introductory ebooks to answer those questions. It's well written and can provide you a good lift
Does this help?
answered Dec 7 '18 at 13:20
seisemanseiseman
1092
$endgroup$
$begingroup$
I'm not a fan of this answer. You are continuing with the assumption that its OK to immediately go into modeling. The data has to be studied first.
$endgroup$
– I_Play_With_Data
Jan 9 at 13:59
add a comment |
$begingroup$
short answer: assess simple models first, and then build more complex ones if necessary.
For a roadmap, I'd go like this:
- get familiar with the different categories of machine learning problems: classification vs regression problems. Your problem falls in the regression category (at least at first sight, you could also reframe it into a classification problem if necessary)
- get familiar with the main corresponding algorithms
- get familiar with the strategies to train / test a model performance
- since you are explicitly asking for online resources, you can have a look at Andrew Ng's machine learning course (especially the first weeks of the course). It's not free, but you can also have a general overview buy looking for his Machine Learning Yearning Book. Another resource coming to my mind is Machine Learning Mastery website (can't remember the guy's name) where you can find some introductory ebooks to answer those questions. It's well written and can provide you a good lift
Does this help?
answered Dec 7 '18 at 13:20
seisemanseiseman
1092
$endgroup$
short answer: assess simple models first, and then build more complex ones if necessary.
For a roadmap, I'd go like this:
- get familiar with the different categories of machine learning problems: classification vs regression problems. Your problem falls in the regression category (at least at first sight, you could also reframe it into a classification problem if necessary)
- get familiar with the main corresponding algorithms
- get familiar with the strategies to train / test a model performance
- since you are explicitly asking for online resources, you can have a look at Andrew Ng's machine learning course (especially the first weeks of the course). It's not free, but you can also have a general overview buy looking for his Machine Learning Yearning Book. Another resource coming to my mind is Machine Learning Mastery website (can't remember the guy's name) where you can find some introductory ebooks to answer those questions. It's well written and can provide you a good lift
Does this help?
answered Dec 7 '18 at 13:20
seisemanseiseman
1092
answered Dec 7 '18 at 13:20
seisemanseiseman
1092
answered Dec 7 '18 at 13:20
seisemanseiseman
1092
answered Dec 7 '18 at 13:20
seisemanseiseman
1092
1092
$begingroup$
I'm not a fan of this answer. You are continuing with the assumption that its OK to immediately go into modeling. The data has to be studied first.
$endgroup$
– I_Play_With_Data
Jan 9 at 13:59
add a comment |
$begingroup$
I'm not a fan of this answer. You are continuing with the assumption that its OK to immediately go into modeling. The data has to be studied first.
$endgroup$
– I_Play_With_Data
Jan 9 at 13:59
$begingroup$
I'm not a fan of this answer. You are continuing with the assumption that its OK to immediately go into modeling. The data has to be studied first.
$endgroup$
– I_Play_With_Data
Jan 9 at 13:59
$begingroup$
I'm not a fan of this answer. You are continuing with the assumption that its OK to immediately go into modeling. The data has to be studied first.
$endgroup$
– I_Play_With_Data
Jan 9 at 13:59
add a comment |
$begingroup$
You are committing some classic mistakes that people make when first entering the world of data science. Your question should not be, "What model do I use?". The question should be, "What can I learn from my data?"
First mistake you're making is jumping straight to modeling - never do that. You need to go through a period of exploratory data analytics (EDA) to help you understand your data. The whole point of an EDA phase is to enable you to ask smarter questions of your project and when you come up with those questions, you will be in a much better position to determine what models you will need. Also, keep in mind that EDA can also serve to answer analytical questions. In your example, your question of spend by card type can probably be answered during a thorough EDA phase.
Second mistake you're making is thinking about the type of models you need without thinking about the questions you want to answer. What's the business case you're trying to solve? What's your hypothesis? What data will you have available when you will run predictions in the future? Do you have a supervised on unsupervised problem on your hands? Etc, etc, etc. These are all things that you need to have some answers to before you can even think about the models that will be used.
Don't forget, this is data science. You need to approach your problems in a methodical, scientific manner in order to truly achieve the results that you seek.
answered Jan 9 at 13:54
I_Play_With_DataI_Play_With_Data
880419
$endgroup$
add a comment |
$begingroup$
You are committing some classic mistakes that people make when first entering the world of data science. Your question should not be, "What model do I use?". The question should be, "What can I learn from my data?"
First mistake you're making is jumping straight to modeling - never do that. You need to go through a period of exploratory data analytics (EDA) to help you understand your data. The whole point of an EDA phase is to enable you to ask smarter questions of your project and when you come up with those questions, you will be in a much better position to determine what models you will need. Also, keep in mind that EDA can also serve to answer analytical questions. In your example, your question of spend by card type can probably be answered during a thorough EDA phase.
Second mistake you're making is thinking about the type of models you need without thinking about the questions you want to answer. What's the business case you're trying to solve? What's your hypothesis? What data will you have available when you will run predictions in the future? Do you have a supervised on unsupervised problem on your hands? Etc, etc, etc. These are all things that you need to have some answers to before you can even think about the models that will be used.
Don't forget, this is data science. You need to approach your problems in a methodical, scientific manner in order to truly achieve the results that you seek.
answered Jan 9 at 13:54
I_Play_With_DataI_Play_With_Data
880419
$endgroup$
add a comment |
$begingroup$
You are committing some classic mistakes that people make when first entering the world of data science. Your question should not be, "What model do I use?". The question should be, "What can I learn from my data?"
First mistake you're making is jumping straight to modeling - never do that. You need to go through a period of exploratory data analytics (EDA) to help you understand your data. The whole point of an EDA phase is to enable you to ask smarter questions of your project and when you come up with those questions, you will be in a much better position to determine what models you will need. Also, keep in mind that EDA can also serve to answer analytical questions. In your example, your question of spend by card type can probably be answered during a thorough EDA phase.
Second mistake you're making is thinking about the type of models you need without thinking about the questions you want to answer. What's the business case you're trying to solve? What's your hypothesis? What data will you have available when you will run predictions in the future? Do you have a supervised on unsupervised problem on your hands? Etc, etc, etc. These are all things that you need to have some answers to before you can even think about the models that will be used.
Don't forget, this is data science. You need to approach your problems in a methodical, scientific manner in order to truly achieve the results that you seek.
answered Jan 9 at 13:54
I_Play_With_DataI_Play_With_Data
880419
$endgroup$
You are committing some classic mistakes that people make when first entering the world of data science. Your question should not be, "What model do I use?". The question should be, "What can I learn from my data?"
First mistake you're making is jumping straight to modeling - never do that. You need to go through a period of exploratory data analytics (EDA) to help you understand your data. The whole point of an EDA phase is to enable you to ask smarter questions of your project and when you come up with those questions, you will be in a much better position to determine what models you will need. Also, keep in mind that EDA can also serve to answer analytical questions. In your example, your question of spend by card type can probably be answered during a thorough EDA phase.
Second mistake you're making is thinking about the type of models you need without thinking about the questions you want to answer. What's the business case you're trying to solve? What's your hypothesis? What data will you have available when you will run predictions in the future? Do you have a supervised on unsupervised problem on your hands? Etc, etc, etc. These are all things that you need to have some answers to before you can even think about the models that will be used.
Don't forget, this is data science. You need to approach your problems in a methodical, scientific manner in order to truly achieve the results that you seek.
answered Jan 9 at 13:54
I_Play_With_DataI_Play_With_Data
880419
answered Jan 9 at 13:54
I_Play_With_DataI_Play_With_Data
880419
answered Jan 9 at 13:54
I_Play_With_DataI_Play_With_Data
880419
answered Jan 9 at 13:54
I_Play_With_DataI_Play_With_Data
880419
880419
add a comment |
add a comment |
Thanks for contributing an answer to Data Science Stack Exchange!
- Please be sure to answer the question. Provide details and share your research!
But avoid …
- Asking for help, clarification, or responding to other answers.
- Making statements based on opinion; back them up with references or personal experience.
Use MathJax to format equations. MathJax reference.
To learn more, see our tips on writing great answers.
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fdatascience.stackexchange.com%2fquestions%2f42266%2fwhen-to-build-separate-models%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
$begingroup$
Usually you don't build separate models unless you really cannot separate cases well. The model depends on your data and your objective, when in doubt try many and see what works better.
$endgroup$
– user2974951
Dec 7 '18 at 7:57