part3.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "JgVF7vWHwEIS"
   },
   "source": [
    "# **PART 3. HANDS-ON DATA VISUALISATION AND ANALYSIS**\n",
    "\n",
    "Below, we will use the data from a social decision-making experiment. Participants had to report the average location of a dot that would randomly appear on a circle. In the second half of the experiment, they were supposedly paired with another participant (they were actually playing with the computer). The computer would be allowed to revise their choice after seeing the participant's decision. The participant saw both the computer's initial and the revised decision. In sporadic intervals, the participants were asked to submit a subjective rating of their performance using a Likert scale.  \n",
    "\n",
    "Previous research has shown that our influence on others affects how we perceive ourselves. We therefore tested whether participants would give themselves a higher performance rating if the computer (who participants thought was a real person) was strongly influenced by 'seeing' their decision, compared to trials with a small difference between the computer's initial and revised choice.\n",
    "\n",
    "An image of a single trial: The white dot shows where the participant thought a blue dot appeared on average. The yellow and orange dots are the computer's initial and revised choices. If the orange dot is close to the white dot (and very far away from the yellow dot), the participant may think that they influenced the computer successfully\n",
    "\n",
    "![Screenshot 2020-11-12 at 00.02.23.png](data:image/png;base64,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)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "MLnNZovhw3pV"
   },
   "source": [
    "We will now try to analyse the data in Python, let's start!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Wdkvw8HdtC-1"
   },
   "source": [
    "## **3.1 ANALYSIS OF CATEGORICAL INDEPENDENT VARIABLES: ANOVA**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### **3.1.1 Data handling with Pandas**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We begin with importing the Excel spreadsheet into Python. As in Part 2, we'll use the pandas package for data analysis!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "NNDm52dXwC9W"
   },
   "outputs": [],
   "source": [
    "import numpy as np \n",
    "import pandas as pd\n",
    "import warnings; warnings.simplefilter('ignore')\n",
    "# Load the data\n",
    "socialDecisionMakingData_ANOVA = pd.read_excel('data/data_BN_task1.xlsx', sheet_name = 'ANOVA')\n",
    "\n",
    "# Before you start any analysis, explore the ANOVA dataset \n",
    "print(socialDecisionMakingData_ANOVA.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "OK, let's utilise the magic of Pandas to explore this dataset!\n",
    "\n",
    "We begin with printing some info about the file:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "socialDecisionMakingData_ANOVA.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Neat! Let's have a look at the actual data using .head():"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# You can have a look at the first five rows using .head()\n",
    "socialDecisionMakingData_ANOVA.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "eWRuXrhDzMYW"
   },
   "outputs": [],
   "source": [
    "# You can look at more (or less) than five rows as well!\n",
    "socialDecisionMakingData_ANOVA.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "tmesZXM7zH3R"
   },
   "outputs": [],
   "source": [
    "# Just as you can look at the first five rows, you can look at \n",
    "# the last five using .tail()\n",
    "socialDecisionMakingData_ANOVA.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Let's just print one variable (== column) from the data sheet\n",
    "socialDecisionMakingData_ANOVA['rating_score'].head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "7VFFl8yu0Quy"
   },
   "source": [
    "### **TASK 6**:\n",
    "\n",
    "1. Try to find a method that will tell you how many responses there are in the 'low', 'medium' and 'high' condition. (**Question**: How do we access a column in Pandas data frame?).\n",
    "\n",
    "2. Try to calculate a grand mean rating score. **Hint**: Remember what you have learned in Part 2 - you can use numpy for this."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "-4mA46ySz2dt"
   },
   "outputs": [],
   "source": [
    "# Type your code below:\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### **3.1.2 Data visualisation with Matplotlib and Seaborn**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Again, we will use seaborn to visualise interesting features of the dataset. Seaborn works great with pandas dataframes, so there's no need to transform the data before we get started."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "e-DhifmBlnJV"
   },
   "source": [
    "**Visualising distribution of data**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "HXyBcjsSkJex"
   },
   "outputs": [],
   "source": [
    "import seaborn as sns\n",
    "\n",
    "# Histogram\n",
    "sns.displot(socialDecisionMakingData_ANOVA, x = \"rating_score\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "yVitWMtGkxvt"
   },
   "outputs": [],
   "source": [
    "# We can choose the size of histogram bins\n",
    "# Compare the two plots below\n",
    "\n",
    "# Plot 1\n",
    "sns.displot(socialDecisionMakingData_ANOVA, x = \"rating_score\", binwidth = 0.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "IraExK7Pk1kT"
   },
   "outputs": [],
   "source": [
    "# Plot 2\n",
    "sns.displot(socialDecisionMakingData_ANOVA, x = \"rating_score\", binwidth = 0.01)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "TWKn2re4l433"
   },
   "source": [
    "Now we can ask the question how this distribution might differ across different levels of the variable 'influence_level'.  \n",
    "We will plot a separate histogram for each of the unique values of 'influence_level'."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "MurYEP5hlk9o",
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "# Feel free to adjust the binwidth so that you can easily read the plot\n",
    "sns.displot(socialDecisionMakingData_ANOVA, x = \"rating_score\", hue = 'influence_level', multiple = 'dodge',binwidth=0.04) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "56ohD7rRnTBs"
   },
   "source": [
    "**Visualising data when you have categorical variables**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the past, you probably created a box plot in SPSS. You can do the same in Python!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "Wru2DhZ4nlJu"
   },
   "outputs": [],
   "source": [
    "# Let's take a closer look at our data\n",
    "# We can specify the type of plot we want by using the argument 'kind'\n",
    "sns.catplot(x=\"influence_level\", y=\"rating_score\",\n",
    "            kind=\"box\", data=socialDecisionMakingData_ANOVA) # This is a so-called box plot "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "_7X3-9fxqvd5"
   },
   "source": [
    "### **TASK 7**:\n",
    "\n",
    "Now try to make a so-called **violin plot** using the same variables we used above (influence_level, rating_score)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "EvWjfI_0oUfO"
   },
   "outputs": [],
   "source": [
    "# Type your code below:\n",
    "\n",
    "# What do the box plot and the violin plot show us?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### **3.1.3 Data preprocessing with Pandas: Removing outliers**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "IqoRBofPrnfW"
   },
   "source": [
    "**Removing outliers**\n",
    "\n",
    "Based on the box plot and the violin plot, there seem to be some data points that significantly differ from other observations.\n",
    "\n",
    "Let's identify and remove the so-called **outliers**."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "2CCZBYie5omk"
   },
   "outputs": [],
   "source": [
    "# First, we want to bring the data into a format where each row corresponds to a single subject.\n",
    "\n",
    "# We also want separate columns for each influence level.\n",
    "\n",
    "# 1. Make a copy of the data\n",
    "df = socialDecisionMakingData_ANOVA\n",
    "\n",
    "# 2. Add subject IDs. The three levels are currently stacked, so we need to repeat the subject ID for each level.\n",
    "df['subject'] = np.tile(np.arange(1,66),3)\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 3. Now we can transform the data such that there is a single row per subject (from long format to wide format).\n",
    "df = df.pivot(index='subject',columns='influence_level',values='rating_score')\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 4. Data transformation\n",
    "# Remember: We define any score above or below 2 SDs as an outlier. \n",
    "# Let's z-score each column and add them as new variables.\n",
    "from scipy.stats import zscore \n",
    "df['zs_high'] = zscore(df['high'])\n",
    "df['zs_medium'] = zscore(df['medium'])\n",
    "df['zs_low'] = zscore(df['low'])\n",
    "df.head(15)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 5. Remove outliers, i.e. subjects where any value is outside of the range we accept.\n",
    "outliers = np.abs(df)>2\n",
    "outlier_mask = np.any(outliers,axis=1)\n",
    "df_clean = df[outlier_mask==False]\n",
    "\n",
    "df_clean.tail(15)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Cool, we have successfully removed a few outliers. \n",
    "\n",
    "Let's have a look at the distribution. For this, we need to transform the data back from wide format (one column per level) to long format (all levels stacked vertically and one column per variable)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Let's only keep the first three columns and reshape from wide to long data (https://anvil.works/blog/tidy-data)\n",
    "subject_ids = np.asarray(df_clean.index)\n",
    "df_clean_long = df_clean[df_clean.columns[:3]].melt(value_name='rating_score')\n",
    "df_clean_long['subject'] = np.tile(subject_ids,3)\n",
    "\n",
    "sns.displot(df_clean_long, x = \"rating_score\", hue = 'influence_level', multiple = 'dodge',binwidth=0.04) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## **3.2 ANOVA WITH STATSMODELS, SCIPY.STATS AND PINGOUIN**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We have a dataset with a categorical IV and a continuous DV. \n",
    "\n",
    "The IV has more than two levels, hence, we want to carry out a **repeated-measures ANOVA**.\n",
    "\n",
    "Do you remember the assumptions?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "uoJPn6nasJVc"
   },
   "source": [
    "**Check assumptions**\n",
    "\n",
    "The assumptions are the following: \n",
    "\n",
    "(a) (approximately) normally distributed dependent variable in all conditions, and \n",
    "\n",
    "(b) sphericity.  \n",
    "\n",
    "We'll use the stats packages Scipy and Pingouin to carry out these tests:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "g4H3rv6ZsS8u"
   },
   "outputs": [],
   "source": [
    "# Check assumptions \n",
    "# Normality: Shapiro-Wilk\n",
    "from scipy.stats import shapiro\n",
    "levels = ['low','medium','high']\n",
    "for lvl in levels:\n",
    "    s,p = shapiro(df_clean[lvl])\n",
    "    print(' test for normality,', lvl ,': W =', str(np.round(s,2)), ', p =', str(np.round(p,4)))\n",
    "\n",
    "# Sphericity\n",
    "from pingouin import sphericity\n",
    "spher,W,chi2,dof,p = sphericity(df_clean[['low','medium','high']])\n",
    "print('\\n test for sphericity: is_spherical: ', spher, ' p =', np.round(p,4))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "6yo_CalVseEi"
   },
   "source": [
    "**Repeated-measures ANOVA**\n",
    "\n",
    "Having visualised the data, removed outliers and checked assumptions, we can now do the main analysis!\n",
    "\n",
    "**Note**: Obviously, we should not do this, as the data isn't normally distributed. ANOVA is quite robust to this violation, but ideally, we would use a non-parametric test instead (see exercise below)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "qSsL0Dyfs3hB"
   },
   "outputs": [],
   "source": [
    "# Repeated Measures ANOVA with statsmodels package\n",
    "from statsmodels.stats.anova import AnovaRM\n",
    "\n",
    "anova = AnovaRM(df_clean_long,depvar='rating_score',subject='subject',within=['influence_level'])\n",
    "results = anova.fit()\n",
    "print(results)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Nice, we have a significant effect! But which difference drives this effect? Let's carry out pairwise comparisons! \n",
    "This is a **post-hoc test**. As this isn't a between-group design, we use a paired t-test, as provided by the scipy package:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Post-hoc tests \n",
    "from scipy.stats import ttest_rel\n",
    "scores = ['low','medium','high']\n",
    "\n",
    "# Let's loop through all possible combinations\n",
    "for ii in range(len(scores)):\n",
    "    for jj in range(ii+1,len(scores)):\n",
    "        lvl1 = df_clean_long['rating_score'][df_clean_long['influence_level']==scores[ii]]\n",
    "        lvl2 = df_clean_long['rating_score'][df_clean_long['influence_level']==scores[jj]]        \n",
    "        t,p = ttest_rel(lvl1,lvl2)        \n",
    "        print('{} vs {}: t({})={}, p={}'.format(scores[ii],scores[jj],str(len(lvl1)-1),str(np.round(t,2)),str(np.round(p,5))))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Cool! We have learned that the difference between low and medium influence is not significant, but the others are!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## **TASK 8**: Non-parametric significance tests for categorical IVs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As pointed out above, we shouldn't have carried out an ANOVA as some of the assumptions were violated. \n",
    "We could have transformed the data, checked for normality again and carried out an ANOVA on the transformed data (provided it would be normally distributed).   \n",
    "Alternatively, we could use non-parametric tests instead, which don't make these distributional assumptions (but are usually less powerful, i.e. worse at detecting an effect).   \n",
    "If you've followed our tutorial carefully, you should have learned enough to do this yourself!  \n",
    "Please carry out:\n",
    "* a Friedman test as a non-parametric alternative to the ANOVA (https://pingouin-stats.org/generated/pingouin.friedman.html) \n",
    "* a Wilcoxon signed-rank test as a non-parametric alternative to the t-test (https://docs.scipy.org/doc/scipy/reference/generated/scipy.stats.wilcoxon.html)\n",
    "\n",
    "**Hint:** The signed-rank test from the scipy.stats package doesn't return the z-statistic, which you would usually need to report together with the p-value. You can obtain an unsigned approximation with the following formula: $$z = ISF(\\frac{p}{2})$$\n",
    "where ISF corresponds to the inverse survival function, which is the inverse of the CDF of a gaussian, i.e. $$ISF(X) = [1-CDF(X)]^{-1}$$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pingouin import friedman\n",
    "\n",
    "# Your code for the Friedman test goes here:\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy.stats import wilcoxon, norm\n",
    "\n",
    "# Your code for the post-hoc tests goes here:\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "XNk4d6m3s-F9"
   },
   "source": [
    "## **3.3 CONTINUOUS VARIABLES: LINEAR REGRESSION AND POST-HOC TESTS WITH SCIKIT-LEARN AND SCIPY.STATS** "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "One of the biggest benefits of coding is that you can automate the boring stuff (there's actually a book on this topic: https://automatetheboringstuff.com/). \n",
    "\n",
    "In statistics, this comes in very handy when you need to repeat the same analysis over and over again.   \n",
    "\n",
    "A good example is the linear regression you sometimes need to conduct (and have maybe already conducted in the Behavioural Neuroscience Core Practical?). Specifically, you need to (i) fit a linear regression model for each subject, (ii) copy & paste the beta value, and (iii) run a t-test on the distribution of these betas to assess significance at the group level.   \n",
    "\n",
    "In Python, you can automate all of this with a simple for loop. Below, we show you how to do this!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "dOXkGi2ntkWI"
   },
   "outputs": [],
   "source": [
    "# Load the data \n",
    "data_forGLM = pd.read_excel('data/data_BN_task1.xlsx', sheet_name = 'GLM')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Let's have a look at the data format\n",
    "data_forGLM.head(15)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# How many entries are there?\n",
    "data_forGLM.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# How many subjects?\n",
    "print('How many subjects? {}'.format(len(np.unique(data_forGLM['subject id']))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "# How many entries for a single subject (e.g., 19)?\n",
    "data_forGLM[data_forGLM['subject id'] == 19]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Cool. \n",
    "\n",
    "The data is in the long format, with separate columns for the DV, the two IVs and a subject identifier. That's all we need. We'll use the linear regression function from scikit-learn, another prominent toolbox for stats and machine learning."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def gen_dmat_and_y(sub_data):\n",
    "    '''\n",
    "    Helper function to set up data \n",
    "    for linear regression\n",
    "    '''\n",
    "    X = np.transpose(np.asarray([zscore(sub_data['influence_level']),zscore(sub_data['error(deg)']),zscore(sub_data['influence_level']*sub_data['error(deg)'])]))\n",
    "    y = np.transpose(np.asarray([sub_data['rating_score']]))\n",
    "    return X,y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "w5poRBCLaJtP"
   },
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LinearRegression\n",
    "from scipy.stats import zscore\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Loop through the participants and get beta values \n",
    "# Create list of subject ids \n",
    "sub_ids = np.unique(data_forGLM['subject id'])\n",
    "\n",
    "# Pre-allocate array for betas (3 columns as there are 2 main effects and an interaction term)\n",
    "betas = np.empty((len(sub_ids),3))\n",
    "\n",
    "# Loop through subjects and estimate betas \n",
    "for ii,sub in enumerate(sub_ids):\n",
    "    # Set-up design matrix \n",
    "    X,y = gen_dmat_and_y(data_forGLM[data_forGLM['subject id']==sub])\n",
    "    \n",
    "    # Fit model \n",
    "    results = LinearRegression().fit(X, y)\n",
    "    \n",
    "    # Store estimated coefficients \n",
    "    betas[ii,:] = results.coef_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "That was quick, no? Let's visualise the distribution of regression weights. We would expect the influence level to be significantly above zero."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Neat, let's have a look at the distribution of betas:\n",
    "plt.figure()\n",
    "_ = plt.boxplot(betas[:,:])\n",
    "plt.xticks(ticks=[1,2,3],labels=['influence lvl','error','interaction'])\n",
    "plt.xlabel('predictor')\n",
    "plt.ylabel(r'$\\beta$ estimate (a.u.)')\n",
    "plt.title('Regression Weights')\n",
    "ax = plt.gca()\n",
    "ax.spines['top'].set_visible(False)\n",
    "ax.spines['right'].set_visible(False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Looking good! Now it's time for a post-hoc test. We'll stick to a simple one-sample t-test on each variable."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Statistical inference on coefficient distributions\n",
    "from scipy.stats import ttest_1samp\n",
    "vars = ['influence lvl','error', 'interaction']\n",
    "for ii,v in enumerate(vars):\n",
    "    t,p = ttest_1samp(betas[:,ii],0)\n",
    "    print('t-test on {var}: t({df})={tscore}, p={pval}'.format(var=v,df=str(len(betas[:,ii])-1),tscore=np.round(t,2),pval=np.round(p,3)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Awesome, that's also consistent with our expectation!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "8-ask0HEc5An"
   },
   "source": [
    "We're now at the end of our little taster session. **Good job!**\n",
    "\n",
    "Most importantly: Don't worry if you feel a little bit overwhelmed. We covered **a lot** of material that is usually taught over several weeks. But this notebook is self-contained and will remain available online. Feel free to revisit it whenever you feel like learning more about Python or brushing up your skills!\n",
    "\n",
    "Also, there is no need to remember every single nitty gritty bit of a programming language. Even experts spend most of their time googling and searching stackoverflow.com for solutions to their problems. That's one of the great benefits of working with computers and being connected to the Internet! A solution is always only a few clicks away.\n",
    "\n",
    "Thanks a lot for your attention!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "FYtC5QIVc5Ao"
   },
   "source": [
    "## RESOURCES\n",
    "Below we have collected a few resources that we think come in handy if you'd like to learn more."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "YFhb-Wdsc5Aq"
   },
   "source": [
    "#### **Learn to code!**\n",
    "1. A complete course https://www.learnpython.org\n",
    "2. Ditto https://www.w3schools.com/python/ \n",
    "3. Advanced course https://automatetheboringstuff.com\n",
    "\n",
    "#### **How to install Python?**\n",
    "1. Just Python\n",
    "https://www.python.org/downloads/\n",
    "2. The Jupyter Notebook (the thing you've been working with today)\n",
    "https://jupyter.org/install\n",
    "3. Anaconda (a collection of useful packages and other software for data scientists)\n",
    "https://www.anaconda.com/distribution/\n",
    "\n",
    "#### **Text editors**\n",
    "Note: If you don't use notebooks, you'll need a text editor to write your Python code. **Don't use Microsoft Word!** You want something that automatically highlights code (and makes it easier to distinguish variables, functions and the like), and provides autocompletion (saves A LOT of time). \n",
    "\n",
    "Below are a few free and popular choices:\n",
    "1. Atom https://atom.io/\n",
    "2. Visual Studio Code https://code.visualstudio.com/\n",
    "3. Sublime https://www.sublimetext.com/\n",
    "\n",
    "All these editors support multiple programming languages.\n",
    "\n",
    "#### **All-in-one solutions**\n",
    "You can also use an all-in-one solution that provides you with an editor, code interpreter (the thing that allows the computer to read and execute Python code) and file managers (a bit like the Matlab software or R-Studio, which you might have heard of). \n",
    "\n",
    "For Python, you can use:\n",
    "\n",
    "1. Spyder (free) https://www.spyder-ide.org/\n",
    "2. Pycharm (free basic and commercial pro version) https://www.jetbrains.com/pycharm/\n",
    "\n",
    "#### **Coding challenges**\n",
    "These challenges are a great way to test your understanding of computer\n",
    "science concepts and improve your logical thinking / problem solving skills.\n",
    "\n",
    "**Pro tip**: Almost every company uses these challenges to assess candidates for software engineering and data science positions.\n",
    "1. Hackerrank https://www.hackerrank.com/ \n",
    "2. Leetcode https://leetcode.com \n",
    "\n",
    "#### **Python for psychologists**\n",
    "You can use Python add-ons to run your own experiments, analyse behavioural, eye-tracking and neuroimaging data\n",
    "and to make beautiful visualisations of your results.\n",
    "\n",
    "Below is a good starting point:\n",
    "https://www.marsja.se/best-python-libraries-psychology/ "
   ]
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [
    "27GJWJouueMT"
   ],
   "name": "intro_to_python_students.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}