From 0bc67f5733101464f0bd75c1d00419cb36e79232 Mon Sep 17 00:00:00 2001 From: Sabine Date: Tue, 18 Nov 2014 23:12:20 +0700 Subject: [PATCH 1/9] draft transcript of our open source masters --- transcripts/sabine-and-fraser-2014.md | 51 +++++++++++++++++++++++++++ 1 file changed, 51 insertions(+) create mode 100644 transcripts/sabine-and-fraser-2014.md diff --git a/transcripts/sabine-and-fraser-2014.md b/transcripts/sabine-and-fraser-2014.md new file mode 100644 index 00000000..46e9ee6a --- /dev/null +++ b/transcripts/sabine-and-fraser-2014.md @@ -0,0 +1,51 @@ +Sabine Crevoisier & Fraser Campbell + +###The Open-Source Masters + +After realising that everything we wanted to learn was online for free in brilliant open source courses, we spent 6 months in Northern Thailand becoming Data Scientists. At the end, we built a website visualising what we learnt: +[Data Science Journey](http://datasciencejourney.com/) + +###Our Backgrounds + +We both originally studied Physics at Cambridge University before becoming an engineer and analytics consultant respectively. However after being introduced to the world of online education and hearing about machine learning at our work we began a journey to become data scientists. + +###Goals & Motivations of the Open Source M.S. + +Having worked as analysts before, we realised there was so much potential in data, if only we had the full technical skillset and statistical grounding to confidently unlock it. Therefore we began studying evenings and weekends to learn. But after realising the length of time this would take, we decided to commit ourselves fully for 6 months mimicking the structure of a masters course. The decision of open source relative to a conventional masters: + 1. Cost + 2. Not as pressurised to attain a certificate due to prior education already achieved + 3. Outstanding quality of teaching and a focus on latest tools and techniques + +###Courses Completed Prior to studying full-time: +* [Computing for Data Analysis](https://www.coursera.org/course/compdata) Coursera course from John Hopkins University - Learn how to program in R and how to use R for effective data analysis +* [Data Analysis](https://www.coursera.org/course/dataanalysis) Coursera course from John Hopkins University - Introduction to data analysis in the R statistical language +* [Machine Learning](https://class.coursera.org/ml-004) Coursera course based upon Andrew Ng's Machine Learning course taught at Stanford University +* [Programming Methodologies](http://see.stanford.edu/errors/default.aspx?aspxerrorpath=/see/courseinfo.aspx) Stanford University's introductory course to functional and object oriented programming using Java +* [Statistics One](https://class.coursera.org/ml-004) Coursera course based on Andrew Conway's course taught at Princeton University aimed at teaching the fundamental concepts in statistics + +###The Data Science Curriculum / May-November 2014 +[Programming Abstractions](http://see.stanford.edu/errors/default.aspx?aspxerrorpath=/see/courseinfo.aspx) Stanford University's follow-up course to programming methodologies, teaching +[Natural Language Processing](https://class.coursera.org/nlangp-001) Coursera course on the application of computational models to text or speech data, based on Michael Collins's course from Columbia University +[Statistics](https://www.udacity.com/course/viewer#!/c-st095) Udacity course +[Stats 110](http://projects.iq.harvard.edu/stat110/home) +[Databases](https://class.coursera.org/db) Coursera course +[Visualisation](http://www.cs171.org/#!index.md) +[Data Science](http://cs109.org/) +[Probabilistic Systems Analysis and Applied Probability](http://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-041-probabilistic-systems-analysis-and-applied-probability-fall-2010/) + +###Projects +* Created an interactive website using D3.js, and did some Bayesian statistical analysis of our study pattern: +[Data Science Journey](http://datasciencejourney.com/) +* Predicted ad click through rates using Kaggle platform. The GB training data enabled us to practise using of an [online learner](http://hunch.net/~vw/) algorithm. [ipython notebook](http://nbviewer.ipython.org/github/fraser-campbell/Machine-Learning-Projects/blob/master/Avazu/Avazu%20Click%20Through%20Rate.ipynb) +* Bike rental prediction using Washington DC data provided by Kaggle. We used linear and log-linear regression methods as well as random forest algorithms to train our model. [ipython notebook](http://nbviewer.ipython.org/github/fraser-campbell/Machine-Learning-Projects/blob/master/Bike%20Sharing/Bike%20Sharing.ipynb) + +###Some books we read or are still reading: +* [Bayesian Methods for Hackers](https://github.com/CamDavidsonPilon/Probabilistic-Programming-and-Bayesian-Methods-for-Hackers) A practical look at MCMC simulations with python libraries +* Bayesian Data Analysis, Andrew Gelman: Book on applying Bayesian Methods +* [Interactive Data Visualisation](http://alignedleft.com/tutorials/d3) Using D3.js to build visualisations in Javascript +* [Data Science with Open Source Tools](http://it-ebooks.info/book/624/) Applications of Data Science in R and Python from a practitioner +* Design for Information, Isabel Meirelles +* [The Elements of Statistical Learning](http://statweb.stanford.edu/~tibs/ElemStatLearn/) This is often cited as the go-to reference for a rigorous mathematical explanation of stats and machine learning methods +* [Programming Abstractions Course Reader](http://web.stanford.edu/class/cs106l/course-reader/full_course_reader.pdf) Textbook following Stanford's C++ Programming course + +**The most valuable resources we found to practise what we learned, and learn more from practitioners was Kaggle From d0212a4972d112ce0d6c7eb39bd4319b36dfcac2 Mon Sep 17 00:00:00 2001 From: Sabine Date: Tue, 18 Nov 2014 23:28:14 +0700 Subject: [PATCH 2/9] updates --- transcripts/sabine-and-fraser-2014.md | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) diff --git a/transcripts/sabine-and-fraser-2014.md b/transcripts/sabine-and-fraser-2014.md index 46e9ee6a..36996096 100644 --- a/transcripts/sabine-and-fraser-2014.md +++ b/transcripts/sabine-and-fraser-2014.md @@ -24,14 +24,14 @@ Having worked as analysts before, we realised there was so much potential in dat * [Statistics One](https://class.coursera.org/ml-004) Coursera course based on Andrew Conway's course taught at Princeton University aimed at teaching the fundamental concepts in statistics ###The Data Science Curriculum / May-November 2014 -[Programming Abstractions](http://see.stanford.edu/errors/default.aspx?aspxerrorpath=/see/courseinfo.aspx) Stanford University's follow-up course to programming methodologies, teaching +[Programming Abstractions](http://see.stanford.edu/errors/default.aspx?aspxerrorpath=/see/courseinfo.aspx) Stanford University's follow-up course to programming methodologies, teaching algorithms, recursion and software design [Natural Language Processing](https://class.coursera.org/nlangp-001) Coursera course on the application of computational models to text or speech data, based on Michael Collins's course from Columbia University -[Statistics](https://www.udacity.com/course/viewer#!/c-st095) Udacity course -[Stats 110](http://projects.iq.harvard.edu/stat110/home) -[Databases](https://class.coursera.org/db) Coursera course -[Visualisation](http://www.cs171.org/#!index.md) -[Data Science](http://cs109.org/) -[Probabilistic Systems Analysis and Applied Probability](http://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-041-probabilistic-systems-analysis-and-applied-probability-fall-2010/) +[Statistics](https://www.udacity.com/course/viewer#!/c-st095) Udacity courses introducing descriptive statistics, study design, t-tests, p-values and ANOVA +[Stats 110](http://projects.iq.harvard.edu/stat110/home) A comprehensive introduction to statistics starting with probability theory. The course has a Bayesian focus and introduces the tools required for understanding statistical inference. +[Databases](https://class.coursera.org/db) Coursera course on Databases, Relational Algebra and the SQL Language. It also contains chapters on XML and NoSQL. +[Visualisation](http://www.cs171.org/#!index.md) Harvard University course on visualising information. It makes extensive use of the D3.js Javascript library to build interactive web visualisations while covering the theory behind human visual cognition. +[Data Science](http://cs109.org/) Harvard University course covering the broad range of skills required to undertake projects as a Data Scientist and communicate your results. The Python development environment was taught and used extensively. +[Probabilistic Systems Analysis and Applied Probability](http://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-041-probabilistic-systems-analysis-and-applied-probability-fall-2010/) We completed the final part of this MIT course, in order to practice statistical inference learnt in Stats 110 above. ###Projects * Created an interactive website using D3.js, and did some Bayesian statistical analysis of our study pattern: @@ -48,4 +48,4 @@ Having worked as analysts before, we realised there was so much potential in dat * [The Elements of Statistical Learning](http://statweb.stanford.edu/~tibs/ElemStatLearn/) This is often cited as the go-to reference for a rigorous mathematical explanation of stats and machine learning methods * [Programming Abstractions Course Reader](http://web.stanford.edu/class/cs106l/course-reader/full_course_reader.pdf) Textbook following Stanford's C++ Programming course -**The most valuable resources we found to practise what we learned, and learn more from practitioners was Kaggle +* **The most valuable resources we found to practise what we learned, and learn more from practitioners was ####Kaggle From a9b2bc42df8c03943ef19477c7a2bb8b6daf37d3 Mon Sep 17 00:00:00 2001 From: Sabine Date: Tue, 18 Nov 2014 23:29:52 +0700 Subject: [PATCH 3/9] minor updates --- transcripts/sabine-and-fraser-2014.md | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/transcripts/sabine-and-fraser-2014.md b/transcripts/sabine-and-fraser-2014.md index 36996096..e3851cb8 100644 --- a/transcripts/sabine-and-fraser-2014.md +++ b/transcripts/sabine-and-fraser-2014.md @@ -12,9 +12,9 @@ We both originally studied Physics at Cambridge University before becoming an en ###Goals & Motivations of the Open Source M.S. Having worked as analysts before, we realised there was so much potential in data, if only we had the full technical skillset and statistical grounding to confidently unlock it. Therefore we began studying evenings and weekends to learn. But after realising the length of time this would take, we decided to commit ourselves fully for 6 months mimicking the structure of a masters course. The decision of open source relative to a conventional masters: - 1. Cost - 2. Not as pressurised to attain a certificate due to prior education already achieved - 3. Outstanding quality of teaching and a focus on latest tools and techniques + * 1. Cost + * 2. Not as pressurised to attain a certificate due to prior education already achieved + * 3. Outstanding quality of teaching and a focus on latest tools and techniques ###Courses Completed Prior to studying full-time: * [Computing for Data Analysis](https://www.coursera.org/course/compdata) Coursera course from John Hopkins University - Learn how to program in R and how to use R for effective data analysis From 31c50ad4229956f141d7f97827310b5acd7dc4c9 Mon Sep 17 00:00:00 2001 From: Sabine Date: Tue, 18 Nov 2014 23:30:19 +0700 Subject: [PATCH 4/9] same --- transcripts/sabine-and-fraser-2014.md | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/transcripts/sabine-and-fraser-2014.md b/transcripts/sabine-and-fraser-2014.md index e3851cb8..dbef1c64 100644 --- a/transcripts/sabine-and-fraser-2014.md +++ b/transcripts/sabine-and-fraser-2014.md @@ -12,9 +12,9 @@ We both originally studied Physics at Cambridge University before becoming an en ###Goals & Motivations of the Open Source M.S. Having worked as analysts before, we realised there was so much potential in data, if only we had the full technical skillset and statistical grounding to confidently unlock it. Therefore we began studying evenings and weekends to learn. But after realising the length of time this would take, we decided to commit ourselves fully for 6 months mimicking the structure of a masters course. The decision of open source relative to a conventional masters: - * 1. Cost - * 2. Not as pressurised to attain a certificate due to prior education already achieved - * 3. Outstanding quality of teaching and a focus on latest tools and techniques +* 1. Cost +* 2. Not as pressurised to attain a certificate due to prior education already achieved +* 3. Outstanding quality of teaching and a focus on latest tools and techniques ###Courses Completed Prior to studying full-time: * [Computing for Data Analysis](https://www.coursera.org/course/compdata) Coursera course from John Hopkins University - Learn how to program in R and how to use R for effective data analysis From 0711e17535c0e47e8d891af032c79d2557d83d9c Mon Sep 17 00:00:00 2001 From: Sabine Date: Tue, 18 Nov 2014 23:32:52 +0700 Subject: [PATCH 5/9] final comment --- transcripts/sabine-and-fraser-2014.md | 2 ++ 1 file changed, 2 insertions(+) diff --git a/transcripts/sabine-and-fraser-2014.md b/transcripts/sabine-and-fraser-2014.md index dbef1c64..7ce700af 100644 --- a/transcripts/sabine-and-fraser-2014.md +++ b/transcripts/sabine-and-fraser-2014.md @@ -49,3 +49,5 @@ Having worked as analysts before, we realised there was so much potential in dat * [Programming Abstractions Course Reader](http://web.stanford.edu/class/cs106l/course-reader/full_course_reader.pdf) Textbook following Stanford's C++ Programming course * **The most valuable resources we found to practise what we learned, and learn more from practitioners was ####Kaggle + +#### The cost of this was 200+ coffees and we will always be grateful to the institutions that made education available to whoever would seize it. \ No newline at end of file From ed4ca4ca7d9726a6d8c592f6449aff952b1cc2f3 Mon Sep 17 00:00:00 2001 From: Sabine Date: Tue, 18 Nov 2014 23:34:37 +0700 Subject: [PATCH 6/9] updated bullet points --- transcripts/sabine-and-fraser-2014.md | 26 +++++++++++++------------- 1 file changed, 13 insertions(+), 13 deletions(-) diff --git a/transcripts/sabine-and-fraser-2014.md b/transcripts/sabine-and-fraser-2014.md index 7ce700af..4599c29b 100644 --- a/transcripts/sabine-and-fraser-2014.md +++ b/transcripts/sabine-and-fraser-2014.md @@ -17,21 +17,21 @@ Having worked as analysts before, we realised there was so much potential in dat * 3. Outstanding quality of teaching and a focus on latest tools and techniques ###Courses Completed Prior to studying full-time: -* [Computing for Data Analysis](https://www.coursera.org/course/compdata) Coursera course from John Hopkins University - Learn how to program in R and how to use R for effective data analysis -* [Data Analysis](https://www.coursera.org/course/dataanalysis) Coursera course from John Hopkins University - Introduction to data analysis in the R statistical language -* [Machine Learning](https://class.coursera.org/ml-004) Coursera course based upon Andrew Ng's Machine Learning course taught at Stanford University -* [Programming Methodologies](http://see.stanford.edu/errors/default.aspx?aspxerrorpath=/see/courseinfo.aspx) Stanford University's introductory course to functional and object oriented programming using Java -* [Statistics One](https://class.coursera.org/ml-004) Coursera course based on Andrew Conway's course taught at Princeton University aimed at teaching the fundamental concepts in statistics +* [Computing for Data Analysis](https://www.coursera.org/course/compdata) Coursera course from John Hopkins University - Learn how to program in R and how to use R for effective data analysis. +* [Data Analysis](https://www.coursera.org/course/dataanalysis) Coursera course from John Hopkins University - Introduction to data analysis in the R statistical language. +* [Machine Learning](https://class.coursera.org/ml-004) Coursera course based upon Andrew Ng's Machine Learning course taught at Stanford University. +* [Programming Methodologies](http://see.stanford.edu/errors/default.aspx?aspxerrorpath=/see/courseinfo.aspx) Stanford University's introductory course to functional and object oriented programming using Java. +* [Statistics One](https://class.coursera.org/ml-004) Coursera course based on Andrew Conway's course taught at Princeton University aimed at teaching the fundamental concepts in statistics. ###The Data Science Curriculum / May-November 2014 -[Programming Abstractions](http://see.stanford.edu/errors/default.aspx?aspxerrorpath=/see/courseinfo.aspx) Stanford University's follow-up course to programming methodologies, teaching algorithms, recursion and software design -[Natural Language Processing](https://class.coursera.org/nlangp-001) Coursera course on the application of computational models to text or speech data, based on Michael Collins's course from Columbia University -[Statistics](https://www.udacity.com/course/viewer#!/c-st095) Udacity courses introducing descriptive statistics, study design, t-tests, p-values and ANOVA -[Stats 110](http://projects.iq.harvard.edu/stat110/home) A comprehensive introduction to statistics starting with probability theory. The course has a Bayesian focus and introduces the tools required for understanding statistical inference. -[Databases](https://class.coursera.org/db) Coursera course on Databases, Relational Algebra and the SQL Language. It also contains chapters on XML and NoSQL. -[Visualisation](http://www.cs171.org/#!index.md) Harvard University course on visualising information. It makes extensive use of the D3.js Javascript library to build interactive web visualisations while covering the theory behind human visual cognition. -[Data Science](http://cs109.org/) Harvard University course covering the broad range of skills required to undertake projects as a Data Scientist and communicate your results. The Python development environment was taught and used extensively. -[Probabilistic Systems Analysis and Applied Probability](http://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-041-probabilistic-systems-analysis-and-applied-probability-fall-2010/) We completed the final part of this MIT course, in order to practice statistical inference learnt in Stats 110 above. +* [Programming Abstractions](http://see.stanford.edu/errors/default.aspx?aspxerrorpath=/see/courseinfo.aspx) Stanford University's follow-up course to programming methodologies, teaching algorithms, recursion and software design +* [Natural Language Processing](https://class.coursera.org/nlangp-001) Coursera course on the application of computational models to text or speech data, based on Michael Collins's course from Columbia University +* [Statistics](https://www.udacity.com/course/viewer#!/c-st095) Udacity courses introducing descriptive statistics, study design, t-tests, p-values and ANOVA. +* [Stats 110](http://projects.iq.harvard.edu/stat110/home) A comprehensive introduction to statistics starting with probability theory. The course has a Bayesian focus and introduces the tools required for understanding statistical inference. +* [Databases](https://class.coursera.org/db) Coursera course on Databases, Relational Algebra and the SQL Language. It also contains chapters on XML and NoSQL. +* [Visualisation](http://www.cs171.org/#!index.md) Harvard University course on visualising information. It makes extensive use of the D3.js Javascript library to build interactive web visualisations while covering the theory behind human visual cognition. +* [Data Science](http://cs109.org/) Harvard University course covering the broad range of skills required to undertake projects as a Data Scientist and communicate your results. The Python development environment was taught and used extensively. +* [Probabilistic Systems Analysis and Applied Probability](http://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-041-probabilistic-systems-analysis-and-applied-probability-fall-2010/) We completed the final part of this MIT course, in order to practice statistical inference learnt in Stats 110 above. ###Projects * Created an interactive website using D3.js, and did some Bayesian statistical analysis of our study pattern: From 7e562a164e4548babdd8cd0139a60ebed4cc4479 Mon Sep 17 00:00:00 2001 From: Sabine Date: Tue, 18 Nov 2014 23:40:55 +0700 Subject: [PATCH 7/9] small updates --- transcripts/sabine-and-fraser-2014.md | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) diff --git a/transcripts/sabine-and-fraser-2014.md b/transcripts/sabine-and-fraser-2014.md index 4599c29b..65c56ae8 100644 --- a/transcripts/sabine-and-fraser-2014.md +++ b/transcripts/sabine-and-fraser-2014.md @@ -19,25 +19,25 @@ Having worked as analysts before, we realised there was so much potential in dat ###Courses Completed Prior to studying full-time: * [Computing for Data Analysis](https://www.coursera.org/course/compdata) Coursera course from John Hopkins University - Learn how to program in R and how to use R for effective data analysis. * [Data Analysis](https://www.coursera.org/course/dataanalysis) Coursera course from John Hopkins University - Introduction to data analysis in the R statistical language. -* [Machine Learning](https://class.coursera.org/ml-004) Coursera course based upon Andrew Ng's Machine Learning course taught at Stanford University. +* [Machine Learning](https://class.coursera.org/ml-004) Coursera course based on Andrew Ng's Machine Learning course taught at Stanford University. * [Programming Methodologies](http://see.stanford.edu/errors/default.aspx?aspxerrorpath=/see/courseinfo.aspx) Stanford University's introductory course to functional and object oriented programming using Java. * [Statistics One](https://class.coursera.org/ml-004) Coursera course based on Andrew Conway's course taught at Princeton University aimed at teaching the fundamental concepts in statistics. ###The Data Science Curriculum / May-November 2014 -* [Programming Abstractions](http://see.stanford.edu/errors/default.aspx?aspxerrorpath=/see/courseinfo.aspx) Stanford University's follow-up course to programming methodologies, teaching algorithms, recursion and software design -* [Natural Language Processing](https://class.coursera.org/nlangp-001) Coursera course on the application of computational models to text or speech data, based on Michael Collins's course from Columbia University +* [Programming Abstractions](http://see.stanford.edu/errors/default.aspx?aspxerrorpath=/see/courseinfo.aspx) Stanford University's follow-up course to programming methodologies, teaching algorithms, recursion and software design. +* [Natural Language Processing](https://class.coursera.org/nlangp-001) Coursera course on the application of computational models to text or speech data, based on Michael Collins's course from Columbia University. * [Statistics](https://www.udacity.com/course/viewer#!/c-st095) Udacity courses introducing descriptive statistics, study design, t-tests, p-values and ANOVA. * [Stats 110](http://projects.iq.harvard.edu/stat110/home) A comprehensive introduction to statistics starting with probability theory. The course has a Bayesian focus and introduces the tools required for understanding statistical inference. -* [Databases](https://class.coursera.org/db) Coursera course on Databases, Relational Algebra and the SQL Language. It also contains chapters on XML and NoSQL. -* [Visualisation](http://www.cs171.org/#!index.md) Harvard University course on visualising information. It makes extensive use of the D3.js Javascript library to build interactive web visualisations while covering the theory behind human visual cognition. +* [Databases](https://class.coursera.org/db) Coursera course on Databases, Relational Algebra and the SQL language. It also contains chapters on XML and NoSQL. +* [Visualisation](http://www.cs171.org/#!index.md) Harvard University course on visualising information. It teaches and makes extensive use of the D3.js Javascript library to build interactive web visualisations while covering the theory behind human visual cognition. * [Data Science](http://cs109.org/) Harvard University course covering the broad range of skills required to undertake projects as a Data Scientist and communicate your results. The Python development environment was taught and used extensively. * [Probabilistic Systems Analysis and Applied Probability](http://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-041-probabilistic-systems-analysis-and-applied-probability-fall-2010/) We completed the final part of this MIT course, in order to practice statistical inference learnt in Stats 110 above. ###Projects * Created an interactive website using D3.js, and did some Bayesian statistical analysis of our study pattern: [Data Science Journey](http://datasciencejourney.com/) -* Predicted ad click through rates using Kaggle platform. The GB training data enabled us to practise using of an [online learner](http://hunch.net/~vw/) algorithm. [ipython notebook](http://nbviewer.ipython.org/github/fraser-campbell/Machine-Learning-Projects/blob/master/Avazu/Avazu%20Click%20Through%20Rate.ipynb) -* Bike rental prediction using Washington DC data provided by Kaggle. We used linear and log-linear regression methods as well as random forest algorithms to train our model. [ipython notebook](http://nbviewer.ipython.org/github/fraser-campbell/Machine-Learning-Projects/blob/master/Bike%20Sharing/Bike%20Sharing.ipynb) +* Predicted ad click through rates using Kaggle platform. The GB training data enabled us to practise using of an [online learner](http://hunch.net/~vw/) algorithm. / [ipython notebook](http://nbviewer.ipython.org/github/fraser-campbell/Machine-Learning-Projects/blob/master/Avazu/Avazu%20Click%20Through%20Rate.ipynb) +* Bike rental prediction using Washington DC data provided by Kaggle. We used linear and log-linear regression methods as well as random forest algorithms to train our model. / [ipython notebook](http://nbviewer.ipython.org/github/fraser-campbell/Machine-Learning-Projects/blob/master/Bike%20Sharing/Bike%20Sharing.ipynb) ###Some books we read or are still reading: * [Bayesian Methods for Hackers](https://github.com/CamDavidsonPilon/Probabilistic-Programming-and-Bayesian-Methods-for-Hackers) A practical look at MCMC simulations with python libraries @@ -48,6 +48,6 @@ Having worked as analysts before, we realised there was so much potential in dat * [The Elements of Statistical Learning](http://statweb.stanford.edu/~tibs/ElemStatLearn/) This is often cited as the go-to reference for a rigorous mathematical explanation of stats and machine learning methods * [Programming Abstractions Course Reader](http://web.stanford.edu/class/cs106l/course-reader/full_course_reader.pdf) Textbook following Stanford's C++ Programming course -* **The most valuable resources we found to practise what we learned, and learn more from practitioners was ####Kaggle +The most valuable resources we found to practise what we learned, and learn more from practitioners was Kaggle #### The cost of this was 200+ coffees and we will always be grateful to the institutions that made education available to whoever would seize it. \ No newline at end of file From d15889cad01177207cbb8f45a75d581c16b32592 Mon Sep 17 00:00:00 2001 From: Sabine Date: Tue, 18 Nov 2014 23:41:58 +0700 Subject: [PATCH 8/9] link to Kaggle --- transcripts/sabine-and-fraser-2014.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/transcripts/sabine-and-fraser-2014.md b/transcripts/sabine-and-fraser-2014.md index 65c56ae8..40fce9c8 100644 --- a/transcripts/sabine-and-fraser-2014.md +++ b/transcripts/sabine-and-fraser-2014.md @@ -48,6 +48,6 @@ Having worked as analysts before, we realised there was so much potential in dat * [The Elements of Statistical Learning](http://statweb.stanford.edu/~tibs/ElemStatLearn/) This is often cited as the go-to reference for a rigorous mathematical explanation of stats and machine learning methods * [Programming Abstractions Course Reader](http://web.stanford.edu/class/cs106l/course-reader/full_course_reader.pdf) Textbook following Stanford's C++ Programming course -The most valuable resources we found to practise what we learned, and learn more from practitioners was Kaggle +The most valuable resources we found to practise what we learned, and learn more from practitioners was [Kaggle](https://www.kaggle.com/) #### The cost of this was 200+ coffees and we will always be grateful to the institutions that made education available to whoever would seize it. \ No newline at end of file From 0ef07d5f35ee28a69cacd654d2313b0c5731b251 Mon Sep 17 00:00:00 2001 From: Sabine Date: Tue, 18 Nov 2014 23:47:01 +0700 Subject: [PATCH 9/9] fraser's updates --- transcripts/sabine-and-fraser-2014.md | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/transcripts/sabine-and-fraser-2014.md b/transcripts/sabine-and-fraser-2014.md index 40fce9c8..dd41fec9 100644 --- a/transcripts/sabine-and-fraser-2014.md +++ b/transcripts/sabine-and-fraser-2014.md @@ -11,12 +11,12 @@ We both originally studied Physics at Cambridge University before becoming an en ###Goals & Motivations of the Open Source M.S. -Having worked as analysts before, we realised there was so much potential in data, if only we had the full technical skillset and statistical grounding to confidently unlock it. Therefore we began studying evenings and weekends to learn. But after realising the length of time this would take, we decided to commit ourselves fully for 6 months mimicking the structure of a masters course. The decision of open source relative to a conventional masters: +Having worked as analysts before, we realised there was so much potential in data, if only we had the full technical skillset and statistical grounding to confidently unlock it. Therefore we began studying evenings and weekends to learn. But after realising the length of time this would take, we decided to commit ourselves fully for 6 months mimicking the structure of a masters course. The decision of open source relative to a conventional masters was based on: * 1. Cost * 2. Not as pressurised to attain a certificate due to prior education already achieved * 3. Outstanding quality of teaching and a focus on latest tools and techniques -###Courses Completed Prior to studying full-time: +###Courses Completed Prior to studying full-time * [Computing for Data Analysis](https://www.coursera.org/course/compdata) Coursera course from John Hopkins University - Learn how to program in R and how to use R for effective data analysis. * [Data Analysis](https://www.coursera.org/course/dataanalysis) Coursera course from John Hopkins University - Introduction to data analysis in the R statistical language. * [Machine Learning](https://class.coursera.org/ml-004) Coursera course based on Andrew Ng's Machine Learning course taught at Stanford University. @@ -36,10 +36,10 @@ Having worked as analysts before, we realised there was so much potential in dat ###Projects * Created an interactive website using D3.js, and did some Bayesian statistical analysis of our study pattern: [Data Science Journey](http://datasciencejourney.com/) -* Predicted ad click through rates using Kaggle platform. The GB training data enabled us to practise using of an [online learner](http://hunch.net/~vw/) algorithm. / [ipython notebook](http://nbviewer.ipython.org/github/fraser-campbell/Machine-Learning-Projects/blob/master/Avazu/Avazu%20Click%20Through%20Rate.ipynb) +* Predicted ad click through rates using Kaggle platform. The GB sized training data enabled us to practise using an [online learner](http://hunch.net/~vw/) algorithm. / [ipython notebook](http://nbviewer.ipython.org/github/fraser-campbell/Machine-Learning-Projects/blob/master/Avazu/Avazu%20Click%20Through%20Rate.ipynb) * Bike rental prediction using Washington DC data provided by Kaggle. We used linear and log-linear regression methods as well as random forest algorithms to train our model. / [ipython notebook](http://nbviewer.ipython.org/github/fraser-campbell/Machine-Learning-Projects/blob/master/Bike%20Sharing/Bike%20Sharing.ipynb) -###Some books we read or are still reading: +###Some books we read or are still reading * [Bayesian Methods for Hackers](https://github.com/CamDavidsonPilon/Probabilistic-Programming-and-Bayesian-Methods-for-Hackers) A practical look at MCMC simulations with python libraries * Bayesian Data Analysis, Andrew Gelman: Book on applying Bayesian Methods * [Interactive Data Visualisation](http://alignedleft.com/tutorials/d3) Using D3.js to build visualisations in Javascript