import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns

%matplotlib inline
sns.set_style("whitegrid")

pokemon = pd.read_csv('pokemon-challenge/pokemon.csv', index_col="#")
combats = pd.read_csv('pokemon-challenge/combats.csv')

pokemon.head()

pokemon.describe()

pokemon.info()

combats.head()

combats.describe()

combats.info()

pk_colors = ['#A8B820',  # Bug,
             '#705848',  # Dark,
             '#7038F8',  # Dragon
             '#F8D030',  # Electric
             '#EE99AC',  # Fairy
             '#C03028',  # Fighting
             '#F08030',  # Fire
             '#A890F0',  # Flying
             '#705898',  # Ghost                    
             '#78C850',  # Grass
             '#E0C068',  # Ground
             '#98D8D8',  # Ice
             '#A8A878',  # Normal
             '#A040A0',  # Poison
             '#F85888',  # Psychic
             '#B8A038',  # Rock
             '#B8B8D0',  # Steel
             '#6890F0',  # Water
             ]

pkt_cnt = pokemon["Type 1"].value_counts(sort=False).sort_index()
pkt_cnt = pd.concat([pkt_cnt, pd.DataFrame(pk_colors,
                                           index=pkt_cnt.index,
                                           columns=["Colors"])], axis=1)
pkt_cnt.sort_values("Type 1", inplace=True)
pkt_cnt_bar = pkt_cnt.plot(kind='barh', y="Type 1", color=pkt_cnt.Colors,
                           legend=False, figsize=(8, 8))
pkt_cnt_bar.set_title("Number of known species\nfor each Pokemon main type",
                      fontsize=16, weight="bold")
pkt_cnt_bar.set_xlabel("Number of species")

pkt = pokemon.pivot_table(index="Type 1",
                          values=["Attack", "Defense", "HP",
                                  "Sp. Atk", "Sp. Def", "Speed"],
                          aggfunc='mean')
pkt.head()

pkt["Total_Power"] = pkt.iloc[:,range(6)].sum(axis=1)
pkt["Color"] = pk_colors

pkt.head()

pkt.sort_values("Total_Power", ascending=True, inplace=True)
pkt_bar = pkt.plot(kind="barh", y="Total_Power", figsize=(8, 8),
                   color=pkt.Color, legend=False)
pkt_bar.set_xlim((350, 575))
pkt_bar.set_xlabel("Mean Total Power")
pkt_bar.set_title("Mean Total Power for each Pokemon Main Type",
                  fontsize=16, weight="bold")

pkl = pokemon.pivot_table(index="Type 1",
                          values=["Legendary"],
                          aggfunc="sum")
pkl.sort_index()
pkl.Legendary = pkl.Legendary.astype(int)
pkl["Total"] = pokemon["Type 1"].value_counts()
pkl["Other"] = pkl.Total - pkl.Legendary
pkl["Ratio"] = pokemon.pivot_table(index="Type 1",
                                   values=["Legendary"],
                                   aggfunc="mean")
pkl["Percent"] = pkl["Ratio"] * 100
pkl["Color"] = pk_colors
pkl.head()

ax = pokemon.Legendary.value_counts().plot.pie(startangle=45,
                                               autopct='%.1f %%',
                                               figsize=(8, 6))
ax.set_title("Proportion of Legendary Pokemons",
             fontsize=16, weight="bold")

pkl_notnull = pkl[pkl.Legendary != 0]
pkl_notnull.head()

l_total = sum(pkl_notnull.Legendary)
ax = pkl_notnull.Legendary.plot(kind="pie", label="",
                                colors=pkl_notnull.Color,
                                autopct='%.1f%%', pctdistance=0.8,
                                explode=pkl_notnull.Legendary / l_total,
                                figsize=(8, 6))
ax.set_title("Legendary Proportions by Pokemon Type",
             fontsize=16, weight="bold")

cols = ["Legendary", "Other"]
fig, axes = plt.subplots(6, 3, figsize=(10, 10))
for i, idx in enumerate(pkl.index):
    ax = axes[i // 3, i % 3]
    ax.pie(pkl[cols].T[idx], labels=cols,
           startangle=30, autopct='%.1f %%')
    ax.set_title(idx, fontdict={"size":"large",
                                "weight":"bold"})

fig.subplots_adjust(wspace=.5, hspace=.5)
plt.suptitle("Proportion of Legendary for each Pokemon Type",
             fontsize=16, weight="bold")

pkg = pokemon.pivot_table(index="Generation",
                          values=["Attack", "Defense", "HP",
                                  "Sp. Atk", "Sp. Def", "Speed"],
                          aggfunc='mean')
pkg.head()

pkg_long = pkg.reset_index().melt(id_vars="Generation")
pkg_long.head(10)

pkg_long.info()

g = sns.FacetGrid(pkg_long, row="variable", size=1.5, aspect=6)
g = g.map(plt.plot, "Generation", "value",
          linestyle='dashed', marker='o')
g.fig.suptitle("Mean Pokemon Stats along Generations",
               fontsize=16, weight="bold", y=1.05)

pkg["Total_Power"] = pkg.sum(axis=1)

ax1 = pkg.Total_Power.plot(kind="line", linestyle="dashed",
                           marker="o", figsize=(10, 4))
pkg["Legendary"] = pokemon.groupby("Generation")[["Legendary"]].sum()
ax2 = pkg.Legendary.plot(kind="line", linestyle="dashed",
                         marker="o", secondary_y=True,
                         ax=ax1)
ax1.set_ylabel("Total Power")
ax2.set_ylabel("# of Legendary")
ax1.legend(loc="upper left")
ax2.legend(loc="upper right")
ax1.set_title("Mean Pokemon Stats along Generations",
               fontsize=16, weight="bold", y=1.05)

ax = pokemon.plot(kind='scatter',
                  x='Attack', y='Defense',
                  alpha = 0.33, color = 'red',
                  figsize=(8, 8))
ax.set_xlabel('Attack')
ax.set_ylabel('Defense')
ax.set_title('Attack vs Defense for every known Pokemon species',
             fontsize=16, weight="bold")

pokemon[(pokemon.Attack < 25) & (pokemon.Defense > 200)]

pokemon[(pokemon.Attack > 175) & (pokemon.Defense < 50)]

ax = pokemon.plot(kind='scatter',
                  x='Attack', y='Defense',
                  alpha = 0.33, color = 'red',
                  figsize=(8, 8))
ax.set_xlabel('Attack')
ax.set_ylabel('Defense')
ax.set_title('Attack vs Defense for every known Pokemon species',
             fontsize=16, weight="bold")
pk1 = pokemon[(pokemon.Attack < 25) & (pokemon.Defense > 200)]
ax.annotate(pk1.Name.iloc[0],
            xy=(pk1.Attack, pk1.Defense), xycoords='data',
            xytext=(30, -10), textcoords='offset points',
            arrowprops=dict(arrowstyle="->", lw=2))
pk2 = pokemon[(pokemon.Attack > 175) & (pokemon.Defense < 50)]
ax.annotate("\n".join(pk2.Name.iloc[0].split()),
            xy=(pk2.Attack, pk2.Defense), xycoords='data',
            xytext=(-100, 0), textcoords='offset points',
            arrowprops=dict(arrowstyle="->", lw=2))

fig, ax = plt.subplots()
fig.set_size_inches(8, 8)
sns.regplot(x="Attack", y="Speed", data=pokemon,
            scatter_kws={'color':'green', 'alpha':0.3},
            line_kws={'color':'red'})
ax.set_title('Attack vs Speed for every known Pokemon species\n+  Regression Line',
             fontsize=16, weight="bold")

g = sns.jointplot(x="Defense", y="Speed",
                 data=pokemon, kind="reg",
                 line_kws={'color':'green'},
                 scatter_kws={'alpha': 0.33})
g.fig.set_size_inches(8, 8)
g.fig.suptitle("Defense vs Speed joinplot\nfor every known Pokemon species",
               fontsize=16, weight="bold", y=1.05)

g = sns.jointplot(x="Defense", y="Speed",
                  data=pokemon, kind="hex")
g.fig.set_size_inches(8, 8)
g.fig.suptitle("Defense vs Speed hexagon joinplot\nfor every known Pokemon species",
               fontsize=16, weight="bold", y=1.05)

pokemon["Total_Power"] = pokemon.iloc[:,range(3, 9)].sum(axis=1)
pokemon.head()

ax = pokemon.boxplot(column='Total_Power',
                     by='Legendary',
                     figsize=(10, 5))
ax.set_ylabel("Total Power")
ax.get_figure().gca().set_title("")
plt.suptitle('Legendary vs non Legendary Total Power',
             fontsize=16, weight="bold")

fig, ax = plt.subplots()
fig.set_size_inches(10, 5)
sns.boxplot(x='Generation', y='Total_Power', data=pokemon)
ax.set_title('Total Power boxplot by Generations',
             fontsize=16, weight="bold")

fig, ax = plt.subplots()
fig.set_size_inches(10, 5)
first_win = sum(combats.First_pokemon == combats.Winner)
second_win = sum(combats.Second_pokemon == combats.Winner)
sns.barplot(x=["First Wins", "Second Wins"],
            y=[first_win, second_win])
ax.set_title('Battle winners count by first attacker order',
             fontsize=16, weight="bold")

first_combats = combats['First_pokemon'].value_counts().sort_index()
second_combats = combats['Second_pokemon'].value_counts().sort_index()
winner_combats = combats['Winner'].value_counts().sort_index()
winner_combats = winner_combats.reindex(first_combats.index, fill_value=0)
battle_stats = pd.concat([first_combats, second_combats, winner_combats], axis=1)
battle_stats["total_combats"] = total_combats = first_combats + second_combats
battle_stats["win_pct"] = winner_combats / total_combats
battle_stats.head()


pokemon["win_pct"] = battle_stats.win_pct
pokemon.head()

g = sns.jointplot(x="Total_Power", y="win_pct",
                  data=pokemon, kind="reg",
                  line_kws={'color':'green'})
g.fig.set_size_inches(8, 8)
g.fig.suptitle("Total Power vs Winning Percentage joinplot\nfor every known Pokemon species",
               fontsize=16, weight="bold", y=1.05)

fig, axes = plt.subplots(2, 3, figsize=(10, 10))

for i, col in enumerate(pokemon.columns[3:9]):
    ax = axes[i // 3, i % 3]
    sns.regplot(data=pokemon, ax=ax,
                y="win_pct", x=col,
                line_kws={"color": "g"},
                scatter_kws={"alpha": 0.2, "color": "r"})
    if (i % 3 > 0):
        ax.set_ylabel("")
        ax.set_yticklabels([])
    ax.set_ylim((0, 1))
fig.suptitle("Winning Percentage vs Pokemon Stats pairgrid\nfor every known Pokemon species",
             fontsize=16, weight="bold", y=0.95)

f,ax = plt.subplots(figsize=(10, 6))
g = sns.heatmap(pokemon[["HP", "Attack", "Defense", "Sp. Atk",
                     "Sp. Def", "Speed", "Total_Power", "win_pct"]].corr(),
            annot=True, linewidths=.5, fmt= '.1f', ax=ax)
plt.suptitle("Pokemon Mean Stats and Wining Ratio\nCorrelation Heatmap",
               fontsize=16, weight="bold", y=1)

fig, axes = plt.subplots(3, 6, figsize=(10, 12))
pkt_win = pokemon[["Type 1", "win_pct"]]
pkt_win = pkt_win.sort_values("Type 1")

grouped = pkt_win.groupby('Type 1')
for i, (key, group) in enumerate(grouped):
    ax = axes[i // 6, i % 6] 
    sns.boxplot(x="Type 1", y="win_pct",
                color="white",
                data=group, ax=ax)
    sns.swarmplot(y='win_pct', color=pk_colors[i],
                  data=group, ax=ax)
    if (i % 6 > 0):
        ax.set_ylabel("")
        ax.set_yticklabels([])
    ax.set_xlabel(key)
    ax.set_ylim((0, 1))

plt.suptitle("Winning Percentage Distribution by Main Pokemon Type",
             fontsize=16, weight="bold", y=0.92)

pokemon[pokemon.win_pct >= 0.95]