Why? Because the fancy data visualization for high-stakes presentations should happen in tools that are the best for it: Tableau<\/a>, Google Data Studio<\/a>, PowerBI, etc… Creating charts and graphs natively in Python should serve only one purpose: to make your data science tasks (e.g. prototyping machine learning models) easier and more intuitive.<\/p>\n\n\n\n So in this tutorial, I’ll focus on how to plot a histogram in Python that’s:<\/p>\n\n\n\n The tool we will use for that is a function in our favorite Python data analytics library — Find the whole code base for this article (in Jupyter Notebook format) here:<\/p>\n\n\n\n Plot histograms in Python (GitHub link)<\/a><\/p>\n\n\n\n Download it from: here<\/a>.<\/p>\n\n\n\n In this article, I assume that you have some basic Python and pandas knowledge.<\/p>\n\n\n\n If you don’t, I recommend starting with these articles:<\/p>\n\n\n\n Also, this is a hands-on tutorial, so it’s the best if you do the coding part with me!<\/p>\n\n\n\n Start with the basics!<\/p>\n\n\n\n What is a histogram and how is it useful?<\/p>\n\n\n\n A histogram shows the number of occurrences of different values in a dataset. At first glance, it is very similar to a bar chart.<\/strong><\/p>\n\n\n\n It looks like this:<\/strong><\/p>\n\n\n\n But a histogram is more than a simple bar chart.<\/strong><\/p>\n\n\n\n Let me give you an example and you’ll see immediately why.<\/p>\n\n\n\n Let’s say that you run a gym and you have 250 clients. For some reason, you want to analyze their height<\/em>s. Good!<\/p>\n\n\n\n You have the individual data points – the height of each and every client in one big Python list:<\/p>\n\n\n\n Note: it’s in centimeters, folks!<\/em><\/p>\n\n\n\n Looking at 250 data points is not very intuitive, is it?<\/p>\n\n\n\n As we’ve discussed in the statistical averages<\/a> and statistical variability<\/a> articles, you have to “compress” these numbers into a few values that are easier to understand yet describe your dataset well enough. These could be:<\/p>\n\n\n\n Based on these values, you can get a pretty good sense of your data\u2026<\/p>\n\n\n\n But if you plot a histogram, too, you can also visualize the distribution of your data points.<\/p>\n\n\n\n For this dataset above, a histogram would look like this:<\/p>\n\n\n\n It’s very visual, very intuitive and tells you even more than the averages and variability measures above. I love it!<\/p>\n\n\n\n You most probably realized that in the height dataset we have ~25-30 unique values. If you simply counted the unique values in the dataset and put that on a bar chart, you would have gotten this:<\/p>\n\n\n\n But when you plot a histogram, there’s one more initial step: these unique values will be grouped into ranges.<\/strong> These ranges are called bins or buckets — and in Python, the default number of bins is As I said: pretty similar to a bar chart — but not the same!<\/p>\n\n\n\n When is this grouping-into-ranges concept useful?<\/strong><\/p>\n\n\n\n For instance when you have way too many unique values in your dataset. (In big data projects, it won’t be ~25-30 as it was in our example… more like 25-30 *million* <\/em>unique values.)<\/p>\n\n\n\n For instance, let’s imagine that you measure the heights of your clients with a laser meter and you store first decimal values, too. Like this:<\/p>\n\n\n\n This is the very same dataset as it was before… only one decimal more accurate.<\/p>\n\n\n\n But because of that tiny difference, now you have not ~25 but ~150 unique values. So if you count the occurrences of each value and put it on a bar chart now, you would get this:<\/p>\n\n\n\n Ouch…<\/p>\n\n\n\n A histogram, though, even in this case, conveniently does the grouping for you. You get values that are close to each other counted and plotted as values of given ranges\/bins:<\/p>\n\n\n\n Beautiful\u2026 but more importantly: useful!<\/p>\n\n\n\n Now that you know the theory, what a histogram is and why it is useful, it’s time to learn how to plot one using Python. There are many Python libraries that can do so:<\/p>\n\n\n\n But I’ll go with the simplest solution: I’ll use the Anyway, the Either way, let’s see how this works!<\/p>\n\n\n\n Note: if you are looking for something eye-catching, check out the seaborn Python dataviz library.<\/em><\/p>\n\n\n\t\t A 100% practical online course. A 6-week simulation of being a junior data scientist at a true-to-life startup.<\/span><\/p> “Solving real problems, getting real experience – just like in a real data science job.”<\/i><\/p>\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t One of the advantages of using the built-in pandas histogram function is that you don’t have to At the very beginning of your project (and of your Jupyter Notebook), run these two lines:<\/p>\n\n\n\n Great! numpy and pandas are imported and ready to use.<\/p>\n\n\n\n And don’t forget to add the:<\/p>\n\n\n\n line, either — so you can plot your charts into your Jupyter Notebook.<\/p>\n\n\n\n As I said, in this tutorial, I assume that you have some basic Python and pandas knowledge. So I also assume that you know how to access your data using Python. (If you don’t, go back to the top of this article and check out the tutorials I linked there.)<\/em><\/p>\n\n\n\n For this tutorial, you don’t have to open any files — I’ve used a random generator to generate the data points of the height data set.<\/p>\n\n\n\n If you want to work with the exact same dataset as I do (and I recommend doing so), copy-paste these lines into a cell of your Jupyter Notebook:<\/p>\n\n\n\n Run them!<\/p>\n\n\n\n For now, you don’t have to know what exactly happened above. (I’ll write a separate article about the In the In the The more complex your data science project is, the more things you should do before you can actually plot a histogram in Python.<\/p>\n\n\n\n Preparing your data is usually more than 80% of the job…<\/p>\n\n\n\n But in this simpler case, you don’t have to worry about data cleaning (removing duplicates, filling empty values, etc.). You just need to turn your Run this line:<\/p>\n\n\n\n Great:<\/p>\n\n\n\n We have the heights of female and male gym members in one big 250-row dataframe.<\/p>\n\n\n\n Before we plot the histogram itself, I wanted to show you how you would plot a line chart and a bar chart that shows the frequency of the different values in the data set\u2026 so you’ll be able to compare the different approaches.<\/p>\n\n\n\n And of course, if you have never plotted anything in pandas before, creating a simpler line chart first can be handy.<\/p>\n\n\n\n To put your data on a chart, just type the If you you’ll get this:<\/p>\n\n\n\n Uhh. Messy.<\/p>\n\n\n\n On the y-axis, you can see the different values of the So let’s tweak this further!<\/p>\n\n\n\n To get what we wanted to get (plot the occurrence of each unique value in the dataset), we have to work a bit more with the original dataset. Let’s add a If you plot the output of this, you’ll get a much nicer line chart:<\/p>\n\n\n\n This is closer to what we wanted\u2026 except that line charts are to show trends. If you want to compare different values, you should use bar charts instead.<\/p>\n\n\n\n To turn your line chart into a bar chart, just add the or:<\/p>\n\n\n\n And of course, you should run this for the This is how you visualize the occurrence of each unique value on a bar chart in Python\u2026<\/p>\n\n\n\n But this is still not a histogram, right!?<\/strong><\/p>\n\n\n\n So…<\/p>\n\n\n\n Once you have your pandas dataframe with the values in it, it’s extremely easy to put that on a histogram.<\/p>\n\n\n\n Type this:<\/p>\n\n\n\n Yepp, compared to the bar chart solution above, the So plotting a histogram (in Python, at least) is definitely a very convenient way to visualize the distribution of your data.<\/strong><\/p>\n\n\n\n If you want a different amount of bins\/buckets than the default 10, you can set that as a parameter. E.g:<\/p>\n\n\n\n Sometimes, you want to plot histograms in Python to compare two different columns of your dataframe.<\/p>\n\n\n\n In that case, it’s handy if you don’t put these histograms next to each other — but on the very same chart.<\/p>\n\n\n\n It can be done with a small modification of the code that we have used in the previous section.<\/p>\n\n\n\n Note: in this version, you called the Anyway, since these histograms are overlapping each other, I recommend setting their transparency to 70% by using the So you can see both charts.<\/p>\n\n\n\n This is it! But you don’t have to\u2026<\/p>\n\n\n\n Anyway, these were the basics. Just use the And don’t stop here, continue with the pandas tutorial episode #5<\/a> where I’ll show you how to plot a scatter plot in pandas<\/a>.<\/p>\n\n\n\n Cheers,<\/em> Cheers,<\/em> Plotting a histogram in Python is easier than you\u2019d think! And in this article, I’ll show you how. I have a strong opinion about visualization in Python, which is: it should be useful and not pretty. Why? Because the fancy data visualization for high-stakes presentations should happen in tools that are the best for it: […]<\/p>\n","protected":false},"author":1,"featured_media":1418,"comment_status":"open","ping_status":"open","sticky":true,"template":"","format":"standard","meta":{"footnotes":""},"categories":[139],"tags":[],"class_list":["post-5091","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-coding-data-science-analytics"],"yoast_head":"\npandas<\/code> — and it’s called .hist()<\/code>… But more about that in the article!<\/p>\n\n\n\nDownload the code base!<\/h2>\n\n\n\n
Before we get started…<\/strong><\/h2>\n\n\n\n
What is a histogram?<\/strong><\/h2>\n\n\n\n
![\"histogram](\"https:\/\/data36.com\/wp-content\/uploads\/2020\/05\/histogram-example-1024x683.png\")
<\/figure>\n\n\n\nheight = [185, 172, 172, 169, 181, 162, 186, 171, 177, 174, 184, 163, 174, 173, 182, 169, 174, 170, 176, 179, 169, 182, 181, 179, 181, 171, 175, 170, 174, 179, 171, 173, 171, 170, 171, 175, 169, 177, 185, 180, 174, 170, 171, 186, 176, 172, 177, 188, 176, 179, 177, 173, 169, 173, 174, 179, 181, 181, 177, 181, 171, 183, 179, 174, 178, 175, 182, 185, 189, 167, 167, 172, 176, 181, 177, 163, 174, 180, 177, 180, 174, 174, 177, 178, 177, 176, 171, 178, 176, 182, 183, 177, 173, 172, 178, 176, 173, 176, 172, 180, 173, 183, 178, 179, 169, 177, 180, 170, 174, 176, 167, 177, 181, 170, 178, 168, 175, 166, 182, 178, 175, 171, 183, 187, 164, 183, 185, 178, 168, 181, 174, 172, 168, 179, 180, 172, 179, 169, 180, 176, 174, 175, 181, 180, 179, 176, 176, 179, 177, 180, 174, 161, 182, 189, 178, 175, 175, 175, 176, 169, 172, 170, 177, 174, 178, 174, 181, 177, 189, 164, 172, 181, 191, 174, 176, 174, 183, 174, 180, 174, 168, 177, 179, 183, 175, 172, 179, 177, 177, 175, 182, 178, 187, 182, 179, 166, 179, 178, 180, 182, 173, 180, 172, 187, 168, 165, 166, 170, 169, 187, 174, 167, 182, 172, 168, 181, 179, 173, 184, 176, 185, 179, 185, 176, 168, 190, 172, 174, 171, 174, 177, 177, 179, 186, 175, 168, 168, 172, 165, 180, 173, 174, 175, 167, 170, 180, 179, 173, 186, 168]
<\/pre>\n\n\n\n175.952<\/code><\/strong><\/li>176<\/code><\/strong><\/li>174<\/code><\/strong><\/li>5.65<\/code><\/strong><\/li>168<\/code><\/strong><\/li>183<\/code><\/strong><\/li><\/ul>\n\n\n\n<\/figure>\n\n\n\nBins and ranges. A histogram is not the same as a bar chart!<\/strong><\/h2>\n\n\n\n
![\"histogram](\"https:\/\/data36.com\/wp-content\/uploads\/2020\/05\/histogram-vs-bar-chart-1024x683.png\")
10<\/code>. So after the grouping, your histogram looks like this:<\/p>\n\n\n\n<\/figure>\n\n\n\nheight = [185.7, 172.3, 172.8, 169.6, 181.2, 162.2, 186.5, 171.4, 177.9, 174.5, 184.8, 163.6, 174.1, 173.7, 182.8, 169.4, 175.0, 170.7, 176.3, 179.5, 169.4, 182.9, 181.4, 179.0, 181.4, 171.9, 175.3, 170.4, 174.4, 179.2, 171.9, 173.6, 171.9, 170.9, 172.0, 175.9, 169.3, 177.4, 186.0, 180.5, 174.8, 170.7, 171.5, 186.2, 176.3, 172.2, 177.1, 188.6, 176.7, 179.7, 177.8, 173.9, 169.1, 173.9, 174.7, 179.5, 181.0, 181.6, 177.7, 181.3, 171.5, 183.5, 179.1, 174.2, 178.9, 175.5, 182.8, 185.1, 189.1, 167.6, 167.3, 173.0, 177.0, 181.3, 177.9, 163.9, 174.2, 181.0, 177.4, 180.6, 174.7, 174.8, 177.1, 178.5, 177.2, 176.7, 172.0, 178.3, 176.7, 182.8, 183.2, 177.1, 173.7, 172.2, 178.5, 176.5, 173.9, 176.3, 172.3, 180.2, 173.3, 183.3, 178.4, 179.6, 169.4, 177.0, 180.4, 170.3, 174.4, 176.2, 167.8, 177.9, 181.1, 170.8, 178.1, 168.1, 175.8, 166.3, 182.7, 178.5, 175.9, 171.3, 183.6, 187.8, 164.9, 183.4, 185.8, 178.0, 168.8, 181.2, 174.9, 172.4, 168.6, 179.3, 180.8, 172.3, 179.1, 169.1, 180.8, 176.3, 174.9, 175.4, 181.2, 180.5, 179.2, 176.8, 176.5, 179.7, 177.4, 180.1, 174.1, 161.4, 182.2, 189.1, 178.6, 175.4, 175.2, 175.3, 176.1, 169.3, 172.9, 170.0, 177.5, 174.2, 179.0, 175.0, 181.9, 177.3, 189.1, 164.6, 172.1, 181.4, 191.2, 174.5, 176.3, 174.6, 184.0, 174.3, 180.1, 174.1, 168.4, 177.9, 179.0, 183.8, 175.3, 172.3, 179.4, 177.4, 177.7, 175.6, 183.0, 178.2, 187.4, 182.7, 180.0, 166.2, 179.6, 178.5, 180.9, 182.3, 173.6, 180.9, 172.6, 187.7, 168.0, 165.4, 166.1, 170.7, 169.3, 187.7, 174.0, 167.9, 182.7, 172.5, 168.6, 181.3, 179.7, 173.4, 184.4, 176.8, 185.7, 179.0, 185.4, 176.7, 168.7, 190.7, 172.7, 174.8, 171.8, 174.8, 177.5, 177.2, 180.0, 186.8, 175.3, 168.6, 168.9, 172.0, 166.0, 181.0, 173.0, 174.1, 176.0, 167.6, 170.8, 180.0, 179.7, 173.3, 186.9, 168.2]<\/pre>\n\n\n\n
![\"bar](\"https:\/\/data36.com\/wp-content\/uploads\/2020\/05\/bar-chart-is-no-historgram.png\")
<\/figure>\n\n\n\n<\/figure>\n\n\n\nHow to plot a histogram in Python (step by step)<\/strong><\/h2>\n\n\n\n
pandas<\/code><\/li>matplotlib<\/code><\/li>seaborn<\/code><\/li>.hist()<\/code> function that’s built into pandas. As I said in the introduction: you don’t have to do anything fancy here… You rather need a histogram that’s useful and informative for you<\/em> — and for your data science tasks.<\/p>\n\n\n\n.hist()<\/code> pandas function is built on top of the original matplotlib solution. (See more info in the documentation<\/a>.) So the result and the visual you’ll get is more or less the same that you’d get by using matplotlib… The syntax will be also similar but a little bit closer to the logic that you got used to in pandas. So in my opinion, it’s better for your learning curve to get familiar with this solution.<\/p>\n\n\n\nThe Junior Data Scientist's First Month<\/h4>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t
Step #1: Import pandas and numpy, and set matplotlib<\/strong><\/h3>\n\n\n\n
import<\/code> any other libraries than the usual: numpy<\/code> and pandas<\/code>.<\/p>\n\n\n\nimport numpy as np\nimport pandas as pd<\/pre>\n\n\n\n
%matplotlib inline<\/pre>\n\n\n\n
![\"import](\"https:\/\/data36.com\/wp-content\/uploads\/2020\/05\/import-numpy-pandas-matplotlib.png\")
<\/figure>\n\n\n\nStep #2: Get the data!<\/strong><\/h3>\n\n\n\n
mu = 168 #mean
sigma = 5 #stddev
sample = 250
np.random.seed(0)
height_f = np.random.normal(mu, sigma, sample).astype(int)<\/pre>\n\n\n\nmu = 176 #mean
sigma = 6 #stddev
sample = 250
np.random.seed(1)
height_m = np.random.normal(mu, sigma, sample).astype(int)<\/pre>\n\n\n\n![\"getting](\"https:\/\/data36.com\/wp-content\/uploads\/2020\/05\/getting-the-data-in-python-histogram-1024x305.png\")
<\/figure>\n\n\n\nnp.random<\/code> function.) Just know that this generated two datasets, with 250 data points in each. And because I fixed the parameter of the random generator (with the np.random.seed()<\/code> line)<\/strong>, you’ll get the very same numpy arrays with the very same data points that I have.<\/p>\n\n\n\nheight_f<\/code> dataset you’ll get 250 height values of female clients of our hypothetical gym.<\/p>\n\n\n\n![\"height](\"https:\/\/data36.com\/wp-content\/uploads\/2020\/05\/height-f-python-histogram-data-generated-1024x462.png\")
<\/figure>\n\n\n\nheight_m<\/code> dataset there are 250 height values of male clients.<\/p>\n\n\n\n![\"height](\"https:\/\/data36.com\/wp-content\/uploads\/2020\/05\/height-m-histogram-data-generated-1024x462.png\")
<\/figure>\n\n\n\nStep #3: Prepare the data!<\/strong><\/h3>\n\n\n\n
height_m<\/code> and height_f<\/code> data into a pandas DataFrame.<\/p>\n\n\n\ngym = pd.DataFrame({'height_f': height_f, 'height_m': height_m})<\/pre>\n\n\n\ngym<\/pre>\n\n\n\n
![\"pandas](\"https:\/\/data36.com\/wp-content\/uploads\/2020\/05\/pandas-dataframe-before-plotting-in-python-1024x482.png\")
<\/figure>\n\n\n\n[OPTIONAL] Basics: Plotting line charts and bar charts in Python using pandas<\/strong><\/h3>\n\n\n\n
.plot()<\/code> function right after the pandas dataframe you want to visualize. By default, .plot()<\/code> returns a line chart.<\/strong><\/p>\n\n\n\nplot()<\/code> the gym<\/code> dataframe as it is:<\/p>\n\n\n\ngym.plot()<\/pre>\n\n\n\n
![\"plot](\"https:\/\/data36.com\/wp-content\/uploads\/2020\/05\/plot-line-chart-python-1024x683.png\")
<\/figure>\n\n\n\nheight_m<\/code> and height_f<\/code> datasets. And the x-axis shows the indexes of the dataframe — which is not very useful in this case.<\/p>\n\n\n\n.groupby()<\/code> with a .count()<\/code> aggregate function. (I wrote more about these in this pandas tutorial<\/a>.)<\/p>\n\n\n\ngym.groupby('height_m').count()<\/pre>\n\n\n\ngym.groupby('height_m').count().plot()<\/pre>\n\n\n\n![\"plot](\"https:\/\/data36.com\/wp-content\/uploads\/2020\/05\/plot-frequency-line-chart-python-1024x683.png\")
bar<\/code> keyword:<\/p>\n\n\n\ngym.groupby('height_m').count().plot.bar()<\/pre>\n\n\n\ngym.groupby('height_m').count().plot(kind='bar')<\/pre>\n\n\n\n![\"plot](\"https:\/\/data36.com\/wp-content\/uploads\/2020\/05\/plot-frequency-bar-chart-1-python-1024x683.png\")
<\/figure>\n\n\n\nheight_f<\/code> dataset, separately:<\/p>\n\n\n\ngym.groupby('height_f').count().plot.bar()<\/pre>\n\n\n\n![\"plot](\"https:\/\/data36.com\/wp-content\/uploads\/2020\/05\/plot-frequency-bar-chart-2-python-1024x683.png\")
<\/figure>\n\n\n\nStep #4: Plot a histogram in Python!<\/strong><\/h3>\n\n\n\n
gym.hist()<\/pre>\n\n\n\n
![\"plot](\"https:\/\/data36.com\/wp-content\/uploads\/2020\/05\/plot-histograms-python-1024x522.png\")
.hist()<\/code> function does a ton of cool things for you, automatically:<\/p>\n\n\n\n
When using .hist()<\/code> there is no need for the initial .groupby()<\/code> function! .hist()<\/code> automatically groups your data into bins. (By default, into 10 bins.)
Note: again, “grouping into bins” is not the same as “grouping by unique values” — as a bin usually contains a range of values.<\/em><\/li>.count()<\/code> function either.)<\/li>gym.hist(bins=20)<\/strong><\/pre>\n\n\n\n
![\"python](\"https:\/\/data36.com\/wp-content\/uploads\/2020\/05\/python-histograms-with-20-bins-1024x717.png\")
<\/figure>\n\n\n\nBonus: Plot your histograms on the same chart!<\/strong><\/h2>\n\n\n\n
gym.plot.hist(bins=20)<\/pre>\n\n\n\n
![\"plot](\"https:\/\/data36.com\/wp-content\/uploads\/2020\/05\/plot-pandas-histogram-1024x683.png\")
<\/figure>\n\n\n\n.hist()<\/code> function from .plot<\/code>.<\/em><\/p>\n\n\n\nalpha<\/code> parameter:<\/p>\n\n\n\ngym.plot.hist(bins=20, alpha=0.7<\/strong>)<\/pre>\n\n\n\n
![\"plot](\"https:\/\/data36.com\/wp-content\/uploads\/2020\/05\/plot-pandas-histogram-transparency-1024x683.png\")
<\/figure>\n\n\n\nConclusion<\/strong><\/h2>\n\n\n\n
Just as I promised: plotting a histogram in Python is easy\u2026 as long as you want to keep it simple. You can make this complicated by adding more parameters to display everything more nicely.<\/p>\n\n\n\n.hist()<\/code> or the .plot.hist()<\/code> functions on the dataframe that contains your data points and you’ll get beautiful histograms that will show you the distribution of your data.<\/p>\n\n\n\n\n
Tomi Mester<\/em><\/strong><\/p>\n\n\n\n
Tomi Mester<\/em><\/strong><\/p>\n","protected":false},"excerpt":{"rendered":"