Twitter | Search | |
Huanfa Chen
78
Tweets
196
Following
136
Followers
Tweets
Huanfa Chen Mar 5
The **factor** data type in R is such a great invention that it help avoid confusing encoders like LabelEncoder, OneHotEncoder, DictVectorizer in sklearn and Python.
Reply Retweet Like
Huanfa Chen Feb 29
In "Data Science for Spatial Systems" tutorials, students asked why not involve spatial autocorrelation in every clustering or regression. Depending on the question, you don't always explicitly consider spatial autocl, as it might be implicitly involved or just not relevant.
Reply Retweet Like
Huanfa Chen Feb 27
UCL Moodle always fails when I need to upload the workshop notebooks... Come back in one hour.
Reply Retweet Like
Huanfa Chen Feb 25
Replying to @andymaclachlan
It's WeChat moment, somewhat similar to Facebook
Reply Retweet Like
Huanfa Chen Feb 24
#marking# In marking, being fair and giving very clear feedback is so important, when you know some students would put their score and feedback on social media and classmates would compare.
Reply Retweet Like
Huanfa Chen Feb 8
My recent blog on Guide to UCL Bartlett Remote Desktop: . Hope it helps.
Reply Retweet Like
Huanfa Chen Feb 6
I checked on UCL wiki, and there is no lecture recording in this venue.
Reply Retweet Like
Huanfa Chen Feb 5
Join us in the CASA seminar!
Reply Retweet Like
Huanfa Chen Feb 5
Replying to @kel_196
Agree... they are also the most time-consuming part
Reply Retweet Like
Huanfa Chen Feb 4
When marking coursework, if figures and tables are proper, it is very likely that the the work is of high quality, although not 100%. Proper figures indicate that the work is not done in one day. BUT, this rule doesn't generalise to academic papers.
Reply Retweet Like
Huanfa Chen Jan 21
Kaggle is a good place to start, with hundreds of real-world datasets and problems.
Reply Retweet Like
Huanfa Chen retweeted
UCL CASA Jan 14
Tomorrow, join us at 5 PM for our first seminar of 2020 - on starting and using data science in arts and culture:
Reply Retweet Like
Huanfa Chen Dec 20
The log transformation on variables makes it really difficult to interpret the model accuracy like R^2 and the weights. In the light of Occam's shaver, should we avoid using log transformation as much as possible?
Reply Retweet Like
Huanfa Chen Dec 17
Gradually I am not that excited by what insights the new data can provide. Rather, I am interested in comparing and bridging classic and emerging methods, and talking about how these methods can be used and **interpreted** correctly.
Reply Retweet Like
Huanfa Chen Dec 16
Obs on fire alarms: Working at CASA for 10 months, 1 fire alarm (5mins). prob=1/215. Been to torrington place 10 times in 10 months, 1 fire alarm (30mins). prob=1/10. Expl: estimated 10 times larger pop in Torrington; more interactions. Conclusion: nice work env in CASA.
Reply Retweet Like
Huanfa Chen Dec 11
Replying to @HuanfaC
In CASA seminar
Reply Retweet Like
Huanfa Chen Dec 11
Ruth talking about 'The social dimension of human navigation'.
Reply Retweet Like
Huanfa Chen Dec 3
Bin is talking about Multilevel Models for MSc students in Quantitative Methods lecture.
Reply Retweet Like
Huanfa Chen Nov 30
CASA blackboard. Copyright Felicia.
Reply Retweet Like
Huanfa Chen Nov 26
I was asked a question in R: can we remove duplicated rows from a df based on the unordered value of two cols? Two rows with (a=0,b=1) and (a=1,b=0) are duplicates and one of them should be removed. I put up a solution here: . Any better solutions?
Reply Retweet Like