Аннотация:One of relevant environmental issues is the determining the presence of various toxicants in the aquatic environment, such as heavy metal ions. The presence of these ions causes disruption of photosynthesis processes and damage to various structures in algae cells, which can induce mass death of phytoplankton and disruption of the balance of aquatic ecosystems. One of the methods to determine the presence of heavy metals in the medium is based on the measurement of chlorophyll a fluorescence induction curves that characterize the state of the photosynthetic apparatus of cells of photosynthetic organisms and alter their shape when exposed to various stress factors. The parameters of the JIP test characterizing the state of individual elements of the photosynthetic apparatus can be calculated from the induction curves. When processing large arrays of such data, it is advisable to use machine learning methods, in particular, the "random forest" classification algorithm. In this work, the green algae Chlorella vulgaris is used as a test organism. The algae cells were incubated for 60 hours, and the fluorescence induction curve was measured once an hour. Toxicants (CdSO4 or K2Cr2O7 at a concentration of 20 or 50 μM) were added at the 17th hour of incubation. From the obtained induction curves, 12 parameters of the JIP test were calculated using the PyPhotoSyn program, which the data array analyzed in this work consists of. The first stage is devoted to consideration of dynamics of JIP-test parameters alteration for the samples with toxicants and without them. It is shown that the impact of cadmium ions in low concentration (20 μM) is almost unnoticeable, because the dynamics of most of parameters for these samples is similar to the control. The impact of chromium ions is noticeable for both concentrations. Fv/Fm, which characterizes the efficiency of photosystem II work, remain about 0.7 for the control data during incubation. PI, which characterizes the overall condition of photosynthetic apparatus, is significantly increased and later decreased. The toxicants inhibit the photosynthetic reaction centers, which is expressed in a decrease of values of these parameters. What is more, this decrease for cadmium samples is quick at the early hours and slow later, unlike chromium samples, for which this decrease is relatively slow. The equality of values of these parameters for cadmium and chromium is detected after 20 hours of toxicant exposure, when it is significantly expressed. ABS/RC characterizes the area of antenna complex per reaction center, the increase of this parameter is usually induced by inhibition of reaction centers. Thus, the inhibition of reaction centers in samples with toxicants induces the increase of ABS/RC and the dynamics of alteration rate of parameters, which is detected earlier, is also noticeable for this parameter. The alteration of Sm, which characterizes the pool of oxidized quinones, are usually induced by alterations in photosynthetic apparatus outside the photosystem II. It is shown that for samples with chromium Sm is decreased and later remains the same, unlike samples with cadmium, for which Sm is decreased and later significantly increased, which can be triggered by chloroplast breath, which is usually observed during stress. Therefore, the cadmium and chromium ions induce the damage in photosynthetic apparatus. However, the influences of these metals are different. The next stage is devoted to use of classifiers. The data which is obtained during the period from 17 to 60 hours of incubation is used to build the random forest classifiers. The JIP-test parameters are the features for the classification. The accuracy of classifier for detection of presence of toxicant is 95% (94% for control samples and 96 % for samples with toxicants). The accuracy of classifier for detection of type of toxicant (cadmium or chromium) is 93%, the best accuracy is for control samples (96%) and samples with chromium (95%), the accuracy for samples with cadmium is lower (87%). The most significant JIP-test parameter for classification is Fv/Fm (the quantum yield of primary photochemistry, which characterizes the efficiency of photosystem II work). To sum it up, the impact of cadmium and chromium ions on algae cells induce the inhibition of reaction centers of photosystem II, which is expressed in a decrease in Fv/Fm and in an increase in ABS/RC. The toxic effect of cadmium is expressed earlier than the effect of chromium. The classifiers for this data could detect the presence and type of toxicant with a high accuracy. These methods for data analysis have prospects of application for estimation of the state of natural habitats.